assuring the security of cloud...
TRANSCRIPT
Assuring the Security of
Cloud Services A framework for evaluating the trustworthiness, resilience and adaptability of
modern business applications that use cloud services and mobile devices
Handbook
James Kavanagh,
National Security Advisor, Microsoft Australia
Disclaimer:
This document has been prepared by Microsoft to provide an overarching framework for compliance and risk
assessment in relation to cloud computing.
This document is provided on an “as is” basis and to the maximum extent permitted by law Microsoft disclaims
all conditions, warranties and guarantees, express or implied, including but not limited to any warranty or
guarantee that the use of the framework set out in this document will not infringe any rights or any warranty
or guarantee of merchantability or fitness for a particular purpose. Before using the framework set out in this
document, you should evaluate its suitability for your organisation. In particular, if you choose to act upon the
output of the framework, then you do so at your own risk.
Apart from any use permitted under the Copyright Act 1968, and the rights explicitly granted below, all rights
are reserved.
Licence: This document is licensed under a Creative Commons Attribution Non-
Commercial 4.0 licence. You are free to copy, distribute and transmit the work
as long as you attribute the authors. You may not use this work for commercial
purposes.
To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc/3.0/au/legalcode.
Contents Executive Summary ................................................................................................................................. 2
Assuring Modern Business Applications ................................................................................................. 4
Modern Apps ...................................................................................................................................... 4
The Problem of Assurance .................................................................................................................. 6
Introducing the SAFE Methodology ........................................................................................................ 8
Evolving security assurance ................................................................................................................ 8
Dimensions of Assurance .................................................................................................................... 8
The Five-Step Process ......................................................................................................................... 9
Step 1 – Understand Strategic Intent ................................................................................................... 18
Strategic Intent & Benefits Case ....................................................................................................... 18
Alternative Options ........................................................................................................................... 19
Context .............................................................................................................................................. 19
Assurance Objectives ........................................................................................................................ 20
Worked Example ............................................................................................................................... 24
Step 2 – Define Requirements .............................................................................................................. 27
Sources of Security Assurance Requirements .................................................................................. 27
Compliance Requirements ................................................................................................................ 28
Security Architecture and Policy Requirements ............................................................................... 33
Worked Example ............................................................................................................................... 35
Step 3 – Verify Claims ........................................................................................................................... 40
Assurance Boundaries ....................................................................................................................... 40
Types of Assurance Claims ................................................................................................................ 41
Worked Example ............................................................................................................................... 47
Step 4 – Assess Risk ............................................................................................................................... 56
Domain Based Risk Assessment ........................................................................................................ 56
Risk Event Catalogue ......................................................................................................................... 59
Threat Modelling ............................................................................................................................... 83
Worked Example ............................................................................................................................. 113
Step 5 – Decide & Plan ........................................................................................................................ 115
Overall Assurance Evaluation ......................................................................................................... 115
Ongoing Governance ...................................................................................................................... 116
Worked Example ............................................................................................................................. 118
1
Figure 1: The five step SAFE methodology for security assurance, progressing from an
organisation’s understanding of strategic intent to definition and verification of requirements,
assessment of risk and preparation of a final executive recommendation
2
Executive Summary
The Challenge The benefits of modern applications and
technology built with cloud services and
mobile devices are well understood:
improved productivity, increased
effectiveness and greater focus on achieving
business outcomes. Governments,
enterprises, small business and consumers are
rapidly adopting these modern applications.
But any enterprise that seriously progresses
towards adopting cloud must diligently
consider issues of compliance and risk. They
need to confirm that they will continue to be
able to satisfy compliance requirements that
may come from external regulators,
legislation, supply contracts and other
sources. They also need to ensure that they
will not face unacceptable risks, even if their
data is held and processed by another
organisation, possibly one based in another
jurisdiction.
A great deal of complexity arises as a result of
the diversity of cloud service providers that
exist in the market. This diversity can be
highly beneficial, enabling an organisation to
find the right combination of services to fit
their needs. But each service will integrate
differently, apply security differently, have
different expectations of privacy and commit
to different contractual terms.
The challenge therefore is to perform due
diligence assessment of compliance and risks,
but in a way that enables direct comparison of
alternative providers. And to do this
assessment quickly and consistently, so that
the business can maintain a focus on
delivering outcomes.
The Big Issues It seems every organisation is thinking about
cloud. They’re considering their future
technology choices and the role that cloud
services and modern applications will play.
This consideration brings into sharp relief a
number of central issues.
Firstly, how will cloud services integrate with
the existing technology, structure, processes
and culture within the organisation?
Important questions here relate to the
capacity of the workforce to adapt and
leverage new technologies and applications,
the process and organisational changes
necessary to realise full benefits and the
technical challenges of identity and access
management, information protection and
monitoring over a hybrid of cloud and on
premise systems.
Secondly, can the organisation satisfy it’s
compliance obligations if it processes or
stores data in the cloud? Concerns about
privacy are well placed with a great deal of
divergence in both the privacy expectations
and practices of cloud service providers. But
compliance considerations in areas like
records management, lawful discovery,
information security and procurement
processes also commonly need to be
assessed.
Finally, will the organisation be exposed to
unacceptable risks in adopting a cloud
service? Although, focus tends to initially turn
to abstract threats based on the most current
threats in the media, there are real and
important risks to consider of a strategic,
operational, technical and compliance nature.
A holistic assessment of risk also needs to
reflect the real risks organisations face in
operating legacy, on premise infrastructure.
In many circumstances, risks in the cloud are
outweighed by risks organisations face by
maintaining business-as-usual approaches.
3
Introducing SAFE The Security Assurance Framework for
Evaluation (SAFE) is a structured, lightweight
methodology for comparative assessment of
compliance and risk in relation to modern
cloud based applications. It is service and
technology agnostic. So it can be used to
compare alternative options that may
comprise such delivery models as on-premise
non-cloud based implementation, private
cloud, hosted cloud, public cloud or any other
combination. It can also be used to compare
multiple alternative cloud provider solutions
on a consistent basis.
It is not designed around any particular sector
or industry. Different tiers of government,
financial services and other sectors all have
unique compliance requirements. So the
SAFE framework is adaptable to any sector,
describing how to elicit and verify compliance
in a formal way. Rather than adopt any one
sector’s compliance or control framework, it
provides a generalised mechanism that an
organisation can apply to their specific
context.
It provides extensive guidance on how to
perform a risk assessment using two
approaches. Firstly, a domain based risk
assessment approach details about fifty
relevant risk events and guides the
organisation to assess both their tolerance
and exposure to those risk events. Secondly,
for more specific risks that require deeper
consideration, the framework provides
formalised threat models that support
definitive risk assessments.
SAFE is designed to be comprehensive but
lightweight, covering conventional assurance
considerations such as confidentiality,
integrity and availability but goes further to
address aspects of privacy, transparency and
adaptability. However, it is designed to be
lightweight, so that the assessment can be
performed rapidly without the need for highly
specialised skills.
Not every organisation will need to perform a
comprehensive compliance assessment or
analyse risks to a deep level. Essentially, SAFE
is a framework for asking questions and
interpreting answers. Its goal is to enable
rapid evaluation and decision making.
It is important to note that SAFE is not a
business case or TCO framework. It focuses
only on aspects of assurance, such as security,
compliance and risk.
Five Steps The SAFE framework consists of five distinct
stages. Firstly, the organisation develops a
clear statement of strategic intent, benefits
and options along with a description of their
assurance objectives. This stage is vital as it
sets the overall context and scope for
everything that follows.
The second stage is about laying out internal
and external requirements. External
requirements may come from legislation,
regulator guidance, codes of conduct, supplier
agreements, and so on. Internal
requirements come from security policies,
existing infrastructure and strategies.
Stage three is about verifying that the cloud
service or application can actually satisfy
these requirements. A combination of service
provider documentation, contract terms and
independent verification can inform this
verification process.
There will always be some unknowns, so the
fourth stage is about performing a holistic risk
assessment across domains of
trustworthiness, resilience and adaptability
considering impacts of a strategic,
operational, compliance or technical nature.
The final step is to combine all of the previous
inputs to provide a recommendation on the
organisation should proceed.
4
Assuring Modern Business Applications
Modern Apps Modern business applications are different.
They combine services delivered from cloud
providers with rich interactive experiences
on devices. They are frequently integrated
with social network services and take
advantage of big data approaches to analyse
and present information. Data within modern
applications is rarely static – it moves from
device to cloud to device, seamlessly crossing
system and geographic boundaries. End users
choose and adapt their own experiences,
combining data from multiple applications,
linking applications to their social network,
consuming content and exploring new
applications at will. Multiple forms of
communication merge on devices as users
seamlessly switch between cloud services that
offer voice, video, short messaging and email.
These modern apps offer the possibility for
employees to be more productive, for
students to learn anywhere, for citizens to
interact with government at any time.
This is a remarkable change from the more
contained, controlled business applications of
the past. The dominant paradigm of the last
ten years of business information systems has
been a focus on enabling silos of functionality,
albeit through web-enabled interfaces. The
paradigm of modern applications is all about
composing multiple discrete services
presented to users through highly adaptable
and engaging experiences on a variety of
devices. Every type of business application,
from email to medical records, from general
ledger to marketing, from transport schedules
to police dispatch, are all moving to this new
model.
The trend to modern applications has arisen
from both the evolving maturity of cloud
services and the increased sophistication of
devices. This is combined with the insistent
expectations of users who now demand
technology that fulfils their needs rather than
adapt their own working style to historic
technological limitations. Cloud services,
social networks, big data analytics and mobile
devices are integrated and mass-customised
to satisfy unique functional needs through
experiences that are tailored for and by users.
Dramatic cost efficiencies are realised through
use of large-scale cloud services and the
productivity benefits of anywhere, anytime
access. More rapid innovation emerges due
to the relative ease of combining multiple
services into useful application experiences.
And new data sources and data analytics
capabilities along with integration to social
networks provide richer insight and more
valuable engagement with customers.
Although this pattern may have begun on
mobile smartphones and small single purpose
utilities and games, it is extending to all forms
of devices and all types of applications. Even
applications that conventionally are
considered more as enabling infrastructure,
like email and storage have modernised into
combined cloud services and on-device
applications.
Figure 2: Modern apps are composed of discrete
cloud services, tailored to unique user needs and
experienced on a variety of mobile devices
5
But new challenges arise with the transition
to modern applications.
Cloud services can be provided at a local or
global scale so data may reside in multiple
geographic locations under different
compliance and regulatory regimes, the
problem often referred to as data sovereignty.
Some cloud services operate with rigorous,
proven security practices while others may
not be so diligent, exposing business and
customer data to the risk of loss. A user’s
work may be disrupted if one of the cloud
services their application relies upon becomes
unavailable. This danger is compounded as
multiple cloud services are combined in a
supply chain or integrated by the application.
Cloud service providers may not apply the
same respect to privacy and appropriate use
of private data as the customer expects,
especially if provision of the cloud service is
subsidised through an advertising or other
‘free’ business model.
Modern devices such as smartphones and
tablets are lit up by interactive experiences
within applications and data sourced from
cloud services. Commonly, modern devices
store cached communications and business
data, so loss of a device can potentially result
in the loss of sensitive data. A device in the
wrong hands or one that is inadequately
protected provides a gateway into the
organisation’s data and business services.
Yet, the demand for smart devices has led to
their proliferation in many forms from the
workplace to the home. These modern
devices blur the distinction between work and
personal use, often holding both corporate
and personal data.
Social network services can be useful both for
individuals as tools of communication and for
organisations as enablers of greater
engagement and connection. Big data
analytical approaches can enable a deeper
level of insight and understanding from vast
stores of data. But social networking and big
data can also be invasive to privacy.
So modern business applications are
composed of widely different cloud services
provided by different parties, from different
locations, under different terms-of-service
and they are experienced by end users on a
variety of different device platforms.
The diversity of choices and challenges is vast.
6
The Problem of Assurance The difficult challenge then for any
organisation adopting modern business
applications is how to validate the security,
privacy, resilience, adaptability and overall
adequacy of the total solution. Organisations
need to have confidence that their data is
protected from unauthorised access and that
their data is only used in a manner that is
consistent with the privacy expectations of
their clients. They need to know that
operational disruptions in cloud or network
service providers will not adversely affect
their own business operations. They need to
be sure that in using these modern
applications, they are still able to comply with
all appropriate regulations pertinent to their
business. More than just technical measures,
they need contractual protections to ensure
service providers say what they do and do
what they say.
And in our dynamic environment of new
opportunities and economic concerns, they
need confidence that their choice today does
not restrict choices they might need to make
tomorrow.
Conventionally, this is a problem within the
domain of information assurance.
Information assurance deals with the
technical and procedural measures designed
to ensure the confidentiality, control,
integrity, authenticity, availability and utility
of information systems. Information
assurance is an established discipline, with
well-defined skills, formal frameworks,
expectations and management
understanding.
But there is a problem applying established
information assurance approaches to modern
applications. Formalised assurance criteria,
standards, compliance requirements, controls
and audit processes have accreted into
sectoral and country-specific tomes. In
exquisite detail, these rigid rule books have
become the ‘bible’ for protection and
management of information. Government
and industry compliance regimes can turn
information assurance into a straitjacket for
the business, rather than an enabler of secure
innovation.
Conventional information assurance
approaches are poorly suited to the reality of
modern business applications. They generally
fail when dealing with modern applications
because of inherent assumptions that no
longer hold, such as:
Uniformity of Control. They generally
assume that all elements of the system
can be controlled to a common level with
similar technical and procedural controls.
But with cloud services provided by
multiple different providers, achieving this
level of control is practically impossible.
Static Data. They assume that data is
relatively static, contained behind
protective perimeters and only interacted
with in very specific ways. But modern
business applications move data in large
quantities between services and devices.
Users interact with data in much more
free-form ways appropriate to their style
of work. Sharing is the default, not
restriction. Data is always in motion.
Predictability. They assume that all
components of the business application
can be predicted in advance. This fails to
recognise the dynamic nature of modern
applications which evolve and adapt to
specific use cases either functionally
improved or enhanced through
integration with new cloud services or
even other apps.
Sovereignty. They frequently fail to
incorporate concerns that only really
surface as a result of the global nature of
cloud services and the new capabilities of
devices. These include for instance issues
of data sovereignty, privacy, data
aggregation and so on.
7
Point-in-time. Most information
assurance approaches are focused on a
point-in-time process such as an audit and
accreditation every few years. There is
typically little capacity to deal with
ongoing continuous improvement,
adaptation and response to changing
circumstances.
These failing assumptions need to be tackled.
But organisations currently operate within an
existing context where established
information assurance practices and
compliance regimes exist. The opportunity is
to evolve these practices to incorporate the
new circumstances of cloud services, social
networking, big data and devices. This is
especially necessary as organisations look to
map out alternative options for how their
communications infrastructure and business
applications will be best delivered into the
future.
There is a need for a security assurance
framework that is relevant to new models of
service & device based business applications,
while reflecting the broad range of security,
privacy and compliance considerations. And
for that assurance framework to be useful for
the rapid evaluation and planning of
organisational strategic choices.
Addressing this need is the principal focus of
the Security Assurance Framework for
Evaluation (SAFE).
8
Introducing the SAFE Methodology
Evolving security assurance Information security assurance commonly
begins from one of two perspectives – risk or
compliance.
A compliance perspective begins by collecting
all of the specific policy and regulatory
requirements, the internal standards and
information security checklists. Then the
organisation works through a series of
checklists, identifying discrepancies in
practice.
A risk perspective begins with enumerating
the assets of an organisation. Vulnerabilities
that may expose these assets to loss are
identified and a list of possible adverse events
along with their impact and probability are
assessed to develop an overall picture of risk
exposure. Then security control and risk
mitigation measures are applied and the final
overall level of risk is assessed.
In reality, most organisations ultimately find
some middle ground between a focus on
compliance and risk. Businesses are often
weighted towards risk management
approaches, while governments typically
favour compliance.
There are a variety of formalised processes to
achieve such assurance including the ISO
27000 series of standards and ISO 31000. The
SAFE methodology builds on these standards
to define a process that any organisation can
use to assure themselves of the security of
the end-to-end modern business application.
Dimensions of Assurance Within the SAFE methodology, security is
treated as a very broad concept
encompassing dimensions of trustworthiness,
resilience and adaptability. This is much
broader than a conventional perspective
limited to confidentiality, availability and
integrity.
Trustworthiness incorporates mechanisms
that ensure confidentiality and integrity of
information but also extends to performance
quality and dependability. A cloud service
that may be highly secure in a conventional
sense but that does exhibit an appropriate
level of transparency, consistent with the
customer’s expectations, would have a limited
assurance of trustworthiness. Similarly a
device or app that is secure against attack but
leaks private information such as user
behaviour and location may have a limited
assurance of trustworthiness.
Figure 3: Dimensions of assurance when using
cloud services
9
Resilience deals with the assurance of
availability and survivability. For a cloud
service, this would incorporate mechanisms
for business continuity, disaster recovery,
automated backup and failover. It also
includes characteristics of services and
devices under exceptional conditions, like
unexpectedly high load.
Finally, adaptability is a key attribute of cloud
solutions but this attribute is perhaps the
least likely to be incorporated into a
conventional assurance model. This is all
about how the organisation can adapt and
reconfigure the business application based on
changing needs and circumstances.
Adaptability may be necessary to cope
with new compliance obligations, a
crisis situation or even to rapidly take
advantage of an emerging opportunity.
Data portability, ease of substitution of
services and choice in suppliers are all
enablers of greater adaptability.
The SAFE methodology supports a
comparative evaluation of an end-to-
end solution in terms of these three
dimensions of assurance.
The Five-Step Process SAFE is designed to guide an
organisation’s evaluation of alternative
options for their implementation of
modern business applications,
principally from the perspective of
security assurance.
It follows a five step process that
progresses from understanding the
intent and context, through
requirement and compliance analysis to
assessing risk and finally planning for
deployment.
Throughout the process, the methodology
incorporates an evaluation of multiple options
but is not prescriptive about which
alternatives they may be. For example, it
could be used to compare email services
provided by a range of different cloud service
providers, to compare an on premise versus a
cloud provider solution or even to compare a
business application implemented with
alternative architectural choices.
The assurance dimensions of trustworthiness,
resilience and adaptability also carry through
the evaluation methodology. Ultimately the
goal is to develop adequate assurance that
the chosen approach delivers on the strategic
intent, appropriate to the organisation’s
internal and external context with an
acceptable level of risk in the dimensions of
trustworthiness, resilience and adaptability.
Figure 4: The 5 Step Process of the Security
Assurance Framework for Evaluation
10
1. Understand Strategic Intent The first stage in any evaluation is to establish
a very clear articulation of the strategic vision
and intended benefits along with a solid
understanding of the organisations internal
and external context.
The strategic intent is important because
security assurance inevitably involves trade-
offs. A particular risk might be demand
treatment with mitigations that adversely
affect functionality or even limit the
achievement of objectives. Typically, the
strategic benefits of modern applications
include faster time to market, ability to enter
into new markets, providing more valuable
services to customers, becoming more
responsive to demand, achieving greater
insight for decision making or achieving
greater efficiencies. There is no one-size-fits-
all. A policing organisation may see the key
strategic intent of a new intelligence
application as enabling rapid response to
emerging incidents and greater insight for
commanding officers. An educational
organisation may view a cloud storage service
combined with a learning app as a strategic
way to streamline and support individualised
student learning.
Security assurance needs to take account of
this strategic intent when assessing required
security functions and appropriately treating
risk. There are often multiple ways to achieve
the desired strategic intent and often the
purpose of evaluation is to help make a choice
between the range of options. Each option
however requires extensive work to assess, so
it is necessary to develop a short-list of viable
alternatives to take through the assurance
methodology.
It is important to reflect that a true assurance
assessment will generally consider both
aspects of maintaining status quo as a
baseline as well as moving to an alternative.
For example, some organisations assess cloud
services and hesitate due to a perceived risk
around data sovereignty. But they already
face risks, sometimes of even greater
proportion, in their existing on-premise
infrastructure, such as legacy system
maintenance and gaps in security practice.
Finally, it’s important to map out the internal
and external context of the organisation,
including how it perceives security risk
management. Context considerations vary
widely between organisations and have
significant bearing on the security assurance
process.
Figure 5: An illustration of some areas of internal
and external context
11
2. Define Requirements
Understanding the organisation context
leads to enumeration of the specific
regulatory and compliance obligations of the
organisation. Most government organisations
need to comply with national or state level
information assurance requirements such as
the Federal Information Security Management
Act (FISMA) requirements in the US or the
Protective Security Policy Framework in
Australia. They may also have specific
legislative requirements depending on their
function, such as legislation relating to health
information privacy, payment card data
protection or national security information.
Specific compliance obligations can really only
be determined by the organisation itself but
the following broad categories generally need
to be considered:
Security & Information Assurance. Public
sector organisations in particular
frequently need to comply with
government mandated security
requirements. These might take the form
of certification, accreditation or formal
evaluation processes. Private sector
organisations can have similar obligations,
such as for example security of credit card
data as specified by the Payment Card
Industry Data Security Standard (PCIDSS)
Privacy. Most organisations face
compliance obligations with regard to
respecting the privacy of customer data.
A distinction is commonly made between
personal information and sensitive
personal information, where the latter
might include credit reporting or medical
data for example. Some organisations
may face additional requirements around
employee privacy. Both cloud services
and devices need to enable these
obligations to be met by the organisation,
either through contractual measures or
capabilities of the service or device.
Records and Audit. Regulated enterprises
and government organisations generally
have specific obligations for formalised
record keeping and audit. In government,
this may extend to enabling freedom of
information requests by citizens or may
relate to the ability of the organisation to
present data to law enforcement on
request.
National Security Information. Central
government, security, intelligence and law
enforcement agencies within government
along with some private sector
organisations in the critical infrastructure
sector may hold national security
classified information. Commonly, the
handling of this information is subject to
special requirements and regulations.
Once the organisation identifies the broadly
relevant regulations, it’s necessary to
enumerate a list of specific compliance
requirements that can be validated and
identify the appropriate mechanism for
compliance verification. For example,
complying with a federal government
information assurance requirement might
require the agency to undertake an
accreditation process for the end-to-end
solution, or it might require only procuring
from an approved panel of evaluated
suppliers, or it may simply be a process of
ensuring contractual protections are in place.
Requirements come from both external and
internal sources. So the final part of this stage
in the evaluation is to identify the key specific
security functional requirements of the
organisation. These are requirements for
integrating the end-to-end solution with
existing infrastructure, aligning appropriately
with the organisations existing information
security management system, or the
capabilities needed to effectively deploy,
manage and monitor the solution. Note that
these requirements are not the business
requirements of the specific solution being
considered, but the security requirements
that would apply broadly to any cloud service
or modern application. For example, this
12
would include acceptable mechanisms of
authentication and access control, required
support for encryption of data across the
network, mechanisms for identity integration,
necessary interfaces for audit and event
reporting and so on.
For the purposes of security assurance and
evaluation, only the key functional
requirements need to be identified as the
more detailed requirements emerge during
planning and implementation phases. The
best way to arrive at these functional
requirements is by attempting to answer
questions like those indicated below.
The aim of mapping these security functional
requirements is to ensure that the overall
solution can integrate well with the
organisation’s existing and future
architecture.
Figure 6: Typical questions to elicit functional security requirements for a
modern app or cloud service
13
3. Verify Claims Each cloud service provider, app developer
and device manufacturer makes claims about
the security of their product or service, and
there are existing mechanisms to validate
those claims. These include:
Evaluation within international or national
information assurance schemes such as
Common Criteria or the US Federal
Information Processing Standards (FIPS).
Certification & Accreditation which
generally requires third party auditing and
validation that the information security
practices conform to a particular
standard, such as ISO 27001.
Audit, either external, internal or
conducted by the end-user organisation
can validate specific claims that may not
be already covered by standards-based
third party accreditation processes.
For cloud services, the most relevant schemes
include ISO 27001 certification, Statement on
Standards for Attestation Engagements No. 16
(SSAE16) audit reports, Payment Card Industry
Data Security Standards (PCI DSS) and the US
FedRamp scheme for the Federal Information
Security Management Act (FISMA)
For devices, formal evaluations are generally
more important. The most widely recognised
evaluations are based on testing for
conformance with Common Criteria
Protection Profiles. However, country-specific
schemes such as the US FIPS and Australian
Evaluated Products List are also important for
specific types of products that contain
cryptographic functions, including
smartphones and tablets.
Other less formal schemes such as the Cloud
Security Alliance (CSA) Registry and the Open
Data Center Alliance (ODCA) Usage Models
can provide a common baseline for the
definition of cloud security capabilities. It’s
important to understand these schemes may
not incorporate validation of claims by a third
party though.
So, the goal of this stage in the evaluation
process is to verify that the solution being
assessed can satisfy both internal and external
requirements. And to ensure that the
assurance claims made by the service provider
or vendor are valid and applicable.
To do this, we first need to consider the
concept of assurance boundaries within an
end-to-end solution. An assurance boundary
demarcates the span of control over security
that any one party has within a solution.
To understand these boundaries, consider an
email application running on a mobile device
connected to a cloud service.
In this example, one assurance boundary can
be drawn around the cloud email service - the
cloud provider controls the security and
operations of the service but has no control
over other parts of the solution. The device
manufacturer controls the security functions
and integrity of the device, so this is another
assurance boundary. A third party may
provide the email application on the device
itself and so this is a third assurance
boundary. Finally, there is a network between
Figure 7: An Example of assurance boundaries
14
the cloud service and the email application.
The provider of this network may vary and will
typically involve multiple parties, but it may
be considered as a single assurance boundary,
albeit one with little assurance.
Typically, each assurance boundary delineates
a set of assurance claims. So a cloud service
provider may claim ISO27001 certification,
FedRAMP Authority to Operate or some other
accreditation. A device provider may claim
their product has been evaluated to FIPS140-
2, to Common Criteria or some other claim.
Every solution will have different assurance
boundaries, delineated by the extent of an
organisations ability to make security
assurance claims. For example, a solution
comprising sensors on devices, a cloud service
and an application on a proprietary device
that is delivered and supported by a single
organisation might comprise of only a single
assurance boundary. Other solutions
comprised, for example, of a cloud service for
storage and a cloud service for email might
have two different assurance boundaries,
even if offered by a single cloud provider if
they cannot make equivalent security claims
over the two services.
Clearly understanding the assurance
boundaries of a solution is essential to
scoping, but it isn’t always necessary to fully
evaluate components within each boundary.
For example, if the strategic intent is to
choose between alternative cloud storage and
backup providers accessible from existing end
user devices, the choice of device may be
irrelevant to the evaluation. So in that case,
there’s no need to verify security claims of the
devices themselves.
Most of the work in this stage comprises of
mapping compliance and functional
requirements to the claims within each
assurance boundary in order to determine
that requirements are adequately addressed.
This mapping process of requirements to
assurance claims is illustrated below.
This mapping of requirements to assurance
claims may be necessary at different levels
depending on the context. For example, one
organisation may have a baseline requirement
of ISO 27001 certification on their suppliers. If
the supplier being evaluated has valid
ISO27001 certification in place, then there’s a
one-to-one mapping of requirement to
assertion. Other organisations may have very
detailed requirements that don’t map easily
to a single assurance claim, and so mapping
must be carried out at the individual level of
specific requirements and controls.
Inevitably, gaps exist between requirements
and assurance claims. When that happens,
further investigation may be required or a risk
assessment on the consequence may be
needed.
Figure 8: Mapping security requirements to
assurance claims
15
4. Assess Risk Modern applications that externalise IT
resources to the cloud can introduce
substantial business change. So even after
specifying and assessing fit to requirements,
holistic risk assessment is required. Although
an organisation may already be facing
significant risk with their on-premise systems,
the adoption of cloud services typically drives
a new awareness of risk and compliance.
The domains of holistic risk assessment
involved are broad, traversing aspects of
strategic, operational, compliance and
technical risk. It can be an onerous, and very
complex task to individually identify threats
and assess their risk to the organisation. So
instead, for the purposes of evaluating
alternative choices, the SAFE methodology
first employs a domain-based comparative
risk assessment.
Effectively this involves a qualitative
assessment of the organisations tolerance and
exposure to a set of risk events. The risk
events are drawn from frameworks such as
the Cloud Security Alliance, the ENISA Security
Risk Assessment model and the ASD Cloud
Considerations. The framework provides a set
of starting questions for each risk event that
assist an organisation in assessing their risk
tolerance and perceived risk exposure.
Tolerance to risk is a statement of how
prepared they are to accept a certain level of
risk. Tolerance to risk is assumed to be
equivalent for a particular risk event across
the variety of options being evaluated. Risk
exposure is conceptually the product of risk
impact and probability, reflected as the effect
of the risk on ability of the organisation to
achieve business objectives. Risk exposure
will vary for each option being considered.
Figure 9: Examples of risk events across each of the assurance domains
and risk impact types
16
As an example, an organisation may be
comparing risk for the option of upgrading an
on-premise email infrastructure versus
migrating to a cloud service provider for
email. Based on their past experience, they
may assess project delivery risk exposure for
the on-premise option as moderate based on
their experience of prior project overruns or
failures. They may assess risk exposure as low
for the cloud-delivered option considering it is
a pre-existing, proven service. For this
particular risk event, the organisation may
have a moderate tolerance for risk based on
their past experience. That is, they could live
with the circumstance where the project
failed to deliver positive benefits but did not
cause damage or loss.
This example is illustrative, but when
conducted across the four key domains of risk
(strategic, operational, compliance and
technical), the approach can provide a useful
high-level comparison of options. It is
consistent with the approach described in ISO
31000 that is often more generally applied
within organisations for enterprise risk
management. It can also help identify specific
areas that require more in-depth risk analysis.
A deeper level of risk analysis in specific areas
may also be justified, looking at particular
threats, events and controls. Formal
approaches to perform this type of analysis
include the creation of threat trees. Threat
trees describe the combination of events that
would need to occur for a specific threat to be
realised. As an example, the following threat
tree in Figure 10 shows a simplistic set of
scenarios for how data contained in a cloud
service could be stolen.
In this example, the primary threat being
considered is for readable customer data to
fall into the hands of a third party. For this to
happen, the threat tree indicates that the
data must be readable, which could happen if
the data is not encrypted, if encryption keys
are known or if the encryption is too weak.
The data must also of course be in the hands
of the third party, which may occur due to an
accidental data spillage, a deliberate hack or a
lawful data access request. The threat tree
can be developed to an arbitrary depth.
The effectiveness of mitigating controls to
prevent each threat is then assessed to
further understand the risk and/or identify
control gaps. Typically threat tree approaches
are much more comprehensive than domain
based comparative evaluation, but these
approaches also require more investment of
time and effort. So generally, they are only
developed for high impact, complex risk
events.
The final step in risk assessment is to make
decisions on appropriate risk acceptance or
treatment options. Some risks will be
deemed acceptable as they are within the risk
Figure 10: Example of a simple threat tree
17
tolerance of the organisation, others may
require the organisation to consider
additional steps to either reduce, avoid or
transfer the risk. Risk reduction measures
may include additional security controls,
further testing or evaluation, contractual
terms, etc. An example of a risk avoidance
measure might include limiting the extent of
functionality enabled or limiting permitted
data in the service to be of a lower sensitivity.
Risk transfer might involve insurance or other
third party arrangements.
Ultimately the aim is to determine if, after
applying optional risk treatments, the overall
risk profile for one or more of the options
being considered is within the overall
organisation risk tolerance.
5. Decide and Plan The final stage is to make a decision about the
best course of action for the organisation by
combining the outcomes of compliance,
functional and risk assessments with the
original strategic intent in mind. This leads to
an overall assurance measure in terms of each
of the three dimensions of assurance:
trustworthiness, resilience and adaptability.
Assurance is not just a point in time activity
though and some planning also has to go into
how decisions made during the assurance
process get incorporated into the engineering
design, how security will be monitored and
how access privileges and other customer
responsibilities for security are enacted.
18
Step 1 – Understand Strategic Intent
Strategic Intent & Benefits Case Any organisation evaluating their options to
implement modern business applications and
technology infrastructure will have some set
of problems or opportunities in mind. For
example, a university may be leveraging cloud
technology to deliver new online learning
opportunities to a broader market, a
government agency might see mobile devices
as essential to modernising service delivery or
an enterprise may see cloud infrastructure as
a mechanism to reduce costs and gain
flexibility. Whatever the scenario, the
business will have an expectation of specific
capabilities being delivered that drive
efficiency, effectiveness and performance
benefits. It’s vital to articulate this strategic
intent and agree on the expected benefits up-
front before undertaking a security assurance
evaluation. This is because decisions about
risk, security controls and levels of compliance
need to be made during the assessment
process in the context of how they relate to
the strategic intent and realisation of benefits.
The strategic intent needs to be clear and
succinct. Consider an organisation in a highly
competitive market that is exploring a cloud
platform to offer dynamic and differentiated
customer services while improving the ability
of their workforce to be productive from
anywhere. Their strategic intent might be:
“To establish a cost-effective technology
platform for marketing and service that
enables us to offer new, dynamic, customer-
centric services while allowing employees to
be highly productive anywhere”
The strategic intent connects with all
subsequent considerations during the
assurance process. Continuing the example,
this organisation may have a higher tolerance
to risk and much lower capacity to absorb cost
overrun. They would favour a service that
assures a high degree of adaptability with the
flexibility to enter new markets, while being
unlikely to accept additional security controls
that limit productivity or efficiency.
A simple model helps to communicate
benefits in different areas of the business.
One way is to classify benefits in a table as
either efficiency, effectiveness or
performance benefits so that the overall
benefits case can be easily understood and
related to by all stakeholders. This is not
intended to be the overall business case, just
a summary to provide context to the
assurance process. Some illustrative benefits
that might be anticipated from a modern
cloud service and device application are
included in the table below.
Business Area Efficiency Benefit Effectiveness Benefit Performance Benefit
Customer Service Reduced time on administrative tasks
Improved service scheduling
Mobile access to customer information
More rapid time-to-market with new services
Customer satisfaction improvement
Increased revenue through profitable new services
Information Technology Infrastructure cost reduction
Ability to scale up and down on demand
More rapid deployment of capabilities
Improved mobile security
Greater agility to align and support the business
Human Resources ICT Staff re-directed to higher value work
Better planning for human resource needs
Increased staff morale by enabling work-at-home
Figure 11: An example of a simple benefits table
19
Alternative Options There are often a number of alternative
architectures or suppliers that can be
deployed to achieve the specific strategic
intent. Each of these options may have very
different assurance characteristics but the aim
is generally to identify the option that best
matches the business needs in a compliant
way with the lowest risk profile.
But organisations also have existing
technology infrastructure so a fair evaluation
often also requires an assessment of the
existing environment. This is because modern
cloud services, mobile devices or applications
may create new risks related to data
sovereignty or third party access, but they
may also reduce existing risks such as legacy
infrastructure dependence and sub-optimal
processes.
Context Understanding the organisation’s context is
crucial to specifying security requirements
and all aspects of risk management. Context
includes considerations of the organisation
objectives, ongoing programs, culture and
attitude towards risk, external regulations,
internal constraints and existing security
practices or perceived vulnerabilities.
External Context Firstly, the external context is considered,
meaning the external environment and
factors outside the organisation that can
affect how the organisation operates or may
operate in the future. These can include
geopolitical, regulatory, social, economic,
market, competition and other factors. As
such a broad realm, it can become an
exhaustive task to catalogue external factors,
but the following questions can be used to
narrow focus around those considerations
that directly impact security assurance.
Context Area Question
Geopolitical Are there political issues in our region that could affect our decisions?
Regulatory Does legislation exist that we need to factor in, including for example, privacy, records and security obligations?
Are there regulator obligations that we need to consider, including for example prudential regulatory requirements?
Are there specific security assurance requirements or policies on our organisation, such as for example federal information security policy?
Are there laws, regulations or requirements in other countries that we operate in that we need to consider?
What legislative or regulatory changes are foreshadowed?
Social Are there community expectations or concerns that we need to factor in, for example privacy and data mining?
Economic What are the prevailing and forecast economic conditions for the organisation?
Market & Competition
What market does the organisation operate within and what are the market dynamics that need to be factored in, such as for example market regulation, consolidation or fragmentation?
What position does the organisation have in the market – leader, follower, niche player?
What activities are our key competitors doing that need to be considered?
External Stakeholders
Who are the key external stakeholders?
Are there views, positions or considerations of these external stakeholders that need to be factored into our decisions?
Figure 12: External Context Questions
20
Internal Context Developing the internal context is about
understanding the organisation’s internal
environment and factors relevant to security
and risk. These factors may include strategic
objectives, structure, key programs, business
functions, risk management practices and
technology infrastructure. Again, the analysis
should be limited to those considerations
relevant to the assurance evaluation.
Context Area Question
Objectives What are the key objectives of the organisation?
What are the key plans, programs and business functions that deliver these organisational objectives?
Are there any internal audits, reports or performance assessments relevant to security assurance?
Structure Which business units will be affected by the solution under consideration?
What resource limitations exist?
How flexible is the organisation to structural change and resource reassignment?
How is the organisation workforce geographically distributed?
Culture Is the workforce culture receptive or resistant to technology innovation?
What is the workforce awareness of risk and security process?
Are there organisation morale or industrial relation issues that may impact on the assurance evaluation?
What is the organisation’s adoption of work-at-home and work-remote practices?
Technology What technology platforms are deployed within the organisation?
How modern is the technology experience of users within the organisation?
Which cloud and mobile device technologies are already used, if any?
Risk Management
What is the organisations current approach around security risk management?
How are risk tolerance decisions made, and by whom?
Figure 13: Internal Context Questions
Assurance Objectives The final step in this first stage is to agree the
assurance objectives for the evaluation.
These can be thought of as the overarching
decision criteria and are expressed as desired
characteristics in the dimensions of
trustworthiness, resilience and adaptability.
Trustworthiness is the foundation of security
assurance, encompassing the ability of the
solution to maintain confidentiality and
integrity. But in the context of a modern
business application, it also incorporates
considerations of privacy, transparency,
predictability and quality. In each of these
areas, the organisation needs to agree desired
characteristics.
For example, think about confidentiality.
Most organisations have some form of
information classification scheme to indicate
sensitivity. If their most sensitively classified
information is to be shared within the solution
being evaluated, then they will certainly
require a high level of assurance around
confidentiality. That assurance expectation
will lead both to security functional
requirements and deep scrutiny about claims
that information can be kept confidential. On
the other hand, if only low classification
materials are being shared, then perhaps a
much lower level of assurance is required and
higher level of risk may be tolerated. It is for
the business organisation to reflect on their
context and strategic intent, then set their
assurance objectives in each area.
Similarly, this approach can be applied to the
dimensions of resilience and adaptability. The
following pages provide a template and
guidance for setting these objectives. It may
be tempting to set the highest bar in each
area, but this is counter-productive as higher
levels of assurance inevitability lead to
additional cost. The aim to keep in mind is to
figure out what the organisation needs as
opposed to what may be merely desirable.
21
Importance
Confidentiality Only public information is being shared so minimal requirement for confidentiality
Internal and sensitive information shared so a reasonable assurance of confidentiality is required.
Moderate tolerance to the risk of an information breach
Highly sensitive or even national security classified information is shared so strict confidentiality is required and very low tolerance
to risks of information breach
Integrity Information stored in the service is not the primary record. Modifications or partial loss would have limited impact
Business decisions are made based on the integrity of
information within the services, but there is a moderate
tolerance for risk of integrity issues in the data.
Information stored is the primary source of record with strict
records preservation regulation. Extensive controls on auditing
and prevention of integrity loss required.
Transparency Limited need-to-know how the service delivers on commitments. Minimal consideration of contract adherence and high risk tolerance for non-compliance
Can take a sensible risk-managed approach but need to know if there are any practices
of the service provider that might be counter to compliance
requirements or expectations
Need explicit and comprehensive visibility of
service provider operations and security practices, most likely
due to regulatory or contractual requirements
Privacy No personally-identifiable data will be stored in the service or on a device
Personal information is stored in the service with policy or
regulations governing how that personal data is collected, used
and managed.
Sensitive personal data (eg. health / financial), is stored with specific compliance regulations
for the management of this sensitive data. Sharing with 3rd
parties is explicitly prohibited
Predictability It has limited consequence for the business if the service changes frequently, adding and removing functionality or discontinuing services even with limited notification
Sensible changes that add beneficial functionality create
few problems, but major changes without notification or
discontinuation of services would have significant impact.
Even minor changes can cause significant disruption to working
practices, so advance notification and release control is
needed for all changes to the service.
Exploitability There would be minimal consequence if vulnerabilities in the device or service were exploited, so limited need for secure development and testing practices
Exploitation of vulnerabilities is undesirable and could impact
the business, so there’s an expectation of sound security
development lifecycle practices
Exploitation of a vulnerability in the service or device could have severe consequences, so formal
evaluation and proven security development lifecycle practices
should be demonstrated
Figure 14: Trustworthiness Assurance Objectives
Trustworthiness
22
Importance
Preparedness Ad-hoc, untested or only partially documented and tested operational continuity processes
Documented processes and procedures for response to
incidents with infrequent or only partial testing of those
procedures
The service provider is highly prepared with documented,
regularly tested, proven procedures to deal with failures
Responsiveness The service provider will respond but place restoration of the service at the lowest priority relative to other customers or services
The service provider will aim to apply resources and restore the service promptly within service
level agreements but other customers or issues may take
priority
The service provider will immediately apply maximum
effective resource to respond to a failure and restore service,
generally with penalties for failure to do so.
Survivability The service fails if even a small part of the overall service is disrupted. Limited or no internal redundancy.
Failover or load-balancing ensure the service survives one-
off failures, but may be susceptible to multiple internal failures or failures in external
services depended upon
The service will survive even a significant amount of disruption
or attack, possibly using multiple internal failovers and alternative
suppliers of external services dependent upon.
Durability It is not significant if the service or device degrades with usage, demonstrating declining performance or functionality over an extended period
Expectation that the service or device does not appreciably
degrade with use over moderate periods of time, but for example,
devices would be replaced in normal asset refresh cycles.
Very limited tolerance for performance degradation over
time. Devices should be especially durable for extreme
environments, extended use and shocks.
Availability Service disruption has limited consequence as users can use alternatives or wait for the service to return. Availability at approximately 98% level is acceptable
The business can cope reasonably with infrequent
disruptions of short duration but has an overall availability expectation above 99.5%
Business tolerance to disruption is very low, with the expectation
that disruption events are extremely rare. Overall service availability needed to be in the
range of 99.95%+.
Load Tolerance Under exceptional load circumstances, it is acceptable for the service performance to significantly degrade to the point of being unusable for periods.
Under exceptional load, the service performance will
degrade but should still be usable
The service must continue to operate with minimal
performance consequences in the case of exceptional load.
Figure 15: Resilience Assurance Objectives
Resilience
23
Importance
Portability Information stored and generated within the service is locked to the service and cannot be exported or used in any other service
Data can be exported from the service but it may be a complex
task to modify its format for import into another service
Tools exist for easy data export in a standardised format that can
then be imported and utilised in another service
Configurability Security settings have minimal configurability, effectively just one-size-fits-all
Generally the range of configuration options are narrow but sufficient for some degree of
customisation to needs.
Extensive configuration options exist enabling the security
settings for access control, information protection and
management to be configured to specific needs.
Flexibility Once the security configuration is set, it is difficult or impossible to make significant changes apart from say, adding/removing users
The security configuration can be modified to suit changing
business needs although it may require manual work
The security configuration is highly flexible and can be reconfigured on demand,
possibly even dynamically in response to changing events
such as attack or high load.
Interoperability The service utilises proprietary or closed interfaces so there is little capability to interoperate with other external services
Some essential interfaces are open and standards based,
especially for authentication and authorisation enabling secure
integration with external services
The service exposes a variety of standard, open interfaces that
enable easy integration with external services.
Substitutability No elements of the service can be substituted
Some basic parts of the service may be substituted, such as for example storage but it is not a normal process that is easily
facilitated
Key elements of the service can be substituted such as
authentication, storage, etc. to satisfy changing business or
assurance needs in a streamlined process.
Extensibility The security configuration cannot be extended beyond the base capabilities within the service
Some extension of the security configuration can be set up, including for example adding
multi-factor authentication, open standards based authorisation or
other key security controls
Substantial extension of the service security can include
capabilities like advanced encryption, custom
authentication and authorisation, and custom monitoring
Figure 16: Adaptability Assurance Objectives
Adaptability
Wo
rked
Exa
mp
le
24
Worked Example
Strategic Intent The New South Wales Government Department of Human Services (a fictitious department for this
example) is investigating options for modernisation of their technology platform for electronic
communications and collaboration. This is in order to improve teamwork, cross-organisation
information sharing and improved ability to serve clients while mobile in the community.
This will include a migration off their legacy email platform currently operated within the
organisation along with the deployment of mobile computing devices that would allow secure,
remote access to communications and client data.
Benefits Case Business Area Efficiency Benefit Effectiveness Benefit Performance Benefit
Client Service Reduced time on administrative tasks involved in communications, information dispatch, etc
Improved service scheduling and coordination
Mobile access to client information enabling more direct service delivery in the community
Improved communications, team work ability and ease of sharing best practices
Improved client satisfaction
Improved coverage of disadvantaged groups through remote and community presence
Information Technology Infrastructure cost reduction across hardware, software and maintenance
Ability to scale up and down on demand
More rapid deployment of new capabilities
Improved security
Redeployed IT resources that can better align to the business needs
Human Resources Ability to redirect human resources to client facing roles
Better planning for human resource needs
Greater flexibility in human resource deployment
Increased staff morale by enabling work-at-home and more direct engagement with the community
Alternative Options There are three principal options being considered:
1. Migrate to an on-premise modern platform for collaboration and communication, owned
and operated by the Department and deploy modern mobile devices for access. It is
anticipated this will be at higher cost, but possibly more specifically configured to the needs
of the Department.
2. Migrate to a dedicated cloud provider service hosted in an Australian data centre and also
deploy modern devices. This is expected to be lower cost than on-premise deployment
3. Migrate to a public cloud service hosted offshore accessed through deployed modern
devices. This is expected to be the lowest cost option, but there is concern as to whether it
introduces data sovereignty and other risks
Wo
rked
Exa
mp
le
25
External Context Context Area Question
Geopolitical No significant geopolitical considerations
Regulatory Existing relevant New South Wales legislation including:
o Privacy and Personal Information Protection Act 1998 (PPIPA): for the handling of personal information by government agencies and contracted service providers
o Health Records and Information Privacy Act 2002 (HRIPA): relevant to information processed by the Department that may relate to healthcare information
o State Records Act 1995 (standards for retention and access to government records). A General Authority (GA35) permits transfer outside of NSW provided an appropriate risk assessment has been made and the Department can maintain compliance with the State Records Act.
o Government Information (Public Access) Act 2009 (GIPAA): Provision that makes government information accessible, except where disclosure would be detrimental
Security assurance requirements exist in the form of the NSW Digital Information Security Policy which requires any cloud service provider to implement management processes for security governance, information security, personnel security and physical security certified to ISO 27001.
The Department operates an Information Security Management System (ISMS) consistent with the NSW Digital Information Security Policy (M2012-15) containing specific policies, standards and procedures of the Department that are broadly aligned to ISO 27001.
Social There is significant sensitivity required to the privacy expectations of clients, as significant amounts of sensitive, personal and financial information is processed by the Department.
Economic The Department is continuing to face budgetary pressures as the demands for client services stretch capacity and ongoing government efficiency dividends need to be delivered annually at 2%+ pa.
Market & Competition
No significant market / competition considerations
External Stakeholders
Key external stakeholders include government, non-government community organisations, contracted community service providers and client groups
Internal Context Context Area Question
Objectives The Department is undergoing a major transformation program to improve service delivery designed around the needs and circumstances of clients. This is driven by the need to do more with less and address inadequacies uncovered in recent government audits.
Structure The solution proposed will affect the entire organisation for communications and collaboration, but will particularly affect the front-line service staff. These staff are already under pressure of heavy workloads and low retention, so disruption to their work needs to be minimised.
The workforce is currently distributed across New South Wales but based primarily in offices. This project will enable staff to be spend a much larger portion of their time away from the office.
Culture The workforce is generally ambivalent to technology if sufficient training has been provided. The introduction of more remote and even work-from-home practices will require significant additional change management. Awareness of security and risk practices are limited, but staff generally follow procedures effectively, particularly in relation to client privacy.
Technology The current technology platform generally in use across the Department for collaboration and communications is out-of-date. End user computing is 90%+ based on office desktop computing.
Client service staff use smartphones for telephone calls and many are familiar with modern consumer apps through their personal use. By policy, the organisation does not currently employ any cloud technology. Some users deliberately or inadvertently breach that policy to use consumer cloud storage and file sharing solutions
Risk Management
The organisation has a generally conservative approach to security risk management, adopting more of a compliance driven approach than one focused on risk-managed adoption of innovation. Ultimately, risk management decisions are made by the Department executive team.
Wo
rked
Exa
mp
le
26
Assurance Objectives
Objective Importance
Trustworthiness Minimal Extreme
A high level of confidentiality and integrity protections are required, although there are no national security or highly damaging information being exchanged. Privacy is a particularly sensitive concern for clients and stakeholders, because some of the data is sensitive personal information. Because of the government nature of services provided, a reasonable level of transparency in how the service provider operates is necessary. The business can tolerate changes in the service as long as there is a reasonable level of predictability that can be factored into plans.
Confidentiality
Integrity
Privacy
Transparency
Predictability
Exploitability
Resilience Minimal Extreme
It is expected that the new communications and collaboration service will have a higher level of resilience than the present on-premise environment which operates at about 99% availability. Although a necessary service for smooth operations of the business, it does not have any life-threatening or huge business impact implications if the service is unexpectedly offline for short periods. Load on the service is generally quite consistent during the working day and even during busy periods does not vary significantly.
Preparedness
Responsiveness
Survivability
Durability
Disruption Tolerance
Load Tolerance
Adaptability Minimal Extreme
As an enterprise scale organisation, the Department has a reasonable level of sophistication in the security configuration required. Because of the ongoing transformation program, it is likely that the security configuration will need to be very flexible to accommodate changing business requirements. However, the business is unlikely to require a high degree of portability of data or ability to substitute or extend security-implementing parts of the service.
Portability
Configurability
Flexibility
Interoperability
Substitutability
Extensibility
27
Step 2 – Define Requirements
Sources of Security Assurance
Requirements Every modern application will have a wide
range of specific functional or technical
requirements that need to be met, but for our
purposes in a security evaluation, our focus is
only on a narrow set of requirements that
relate specifically to two areas: compliance
requirements and security functional
requirements.
Compliance requirements generally arise from
external legislation, regulations, policies,
standards and contracts. Some examples
would include privacy legislation, government
security assurance standards, payment card
industry rules and outsourcing contract terms.
Often but not always, external compliance
requirements are given effect by internal
policies which define more specifically how an
external compliance requirement is to be
enforced by the organisation.
In the case of a cloud service, most of the
compliance requirements need to be satisfied
by the service consumer organisation itself,
contracting appropriately with the service
provider. The reality is that compliance
requirements can’t simply be transferred
without accountability to the service provider.
For example, consider some of the typical
requirements for privacy compliance. The
service consumer organisation needs to
ensure that they are compliant with all of the
provisions for collection, storage, use,
correction, destruction, etc. They may
formally contract the service provider under
terms that ensure the information is stored
securely, but the compliance obligation rests
with the service consumer, not the cloud
service provider (or perhaps as well as the
cloud service provider).
Security functional requirements relate to
specifications that should be met by the
solution in areas like identity & access control,
management & monitoring and information
protection. They arise from the fact that
organisation already have existing
infrastructure that need to be integrated with.
For example, if the organisation has already
deployed multi-factor authentication tokens
to all users, it’s sensible that the cloud service,
device or app being evaluated should be able
to support those tokens. Likewise, if the
organisation already uses a set of tools for
managing their IT systems, it may be
necessary for the cloud service or device to
actually work with those tools. Requirements
may translate into security controls, but the
intent here is not to develop an exhaustive list
of required controls, but rather to identify
needs specific to the organisation.
Figure 17: Sources of external and internal assurance requirements
28
Compliance Requirements A consideration of compliance requirements
is necessary to avoid breaches of any law,
regulatory, statutory or policy obligation.
These requirements will often be specific to
the jurisdiction, sector and operations of the
organisation. Although compliance
obligations around information management
generally exist regardless of the use of cloud,
the questions of data location and control
that arise in cloud computing often bring an
awareness of compliance obligations to the
fore. Advice on specific legal requirements
should always be sought from the
organisation’s legal advisers, or qualified legal
practitioners. The following sections outline
some of the considerations an organisation
should include but does not constitute legal
advice.
Identifying and documenting compliance
requirements can become onerous, so this
evaluation methodology proposes a
streamlined approach to identify the most
relevant compliance requirements, rather
than an exhaustive list. The starting point
should be to understand the organisations
current practices around compliance and then
expand to survey other potential compliance
obligations relevant to the solution under
evaluation.
For each potential compliance requirement,
there are a few key questions to ask:
1. What are the source and specifics of the
compliance requirement? This is in
contrast to it’s interpretation in policy or
current practice. To understand the basis
of this question, consider the case of
privacy and data location considerations.
Many organisations initially think that
some legislation or regulation exists that
prevents private data from being located
offshore. Deeper investigation of the
actual compliance requirement often
reveals that the legal obligation does not
prohibit data location offshore, but current
practice or policy within the organisation is
to keep data onshore. Current practice or
policy is within the scope of control of the
organisation and so it can be adapted. It is
also necessary to understand if the
requirement is mandatory, recommended
or best practice.
2. How does the organisation currently
satisfy the compliance requirement? It is
often an incorrect assumption that current
practices are fully compliant, and
organisations may find that migration to
modern applications can actually improve
compliance and security. Understanding
how requirements are currently satisfied
can guide the organisation in defining
appropriate requirements for the modern
application or identify gaps in current
practice. This also helps to define how
compliance should be verified.
3. Who is the appropriate authority for
compliance? Typically an organisation,
board or individual is accountable for
compliance to a specific requirement. For
example, the CEO or CFO may be
accountable for corporate governance
obligations, the Chief Security Officer may
be accountable for information assurance
requirements, or an external party like a
government security agency may be
accountable. Sometimes the accountable
party has the flexibility to choose not to be
compliant on certain requirements, as long
as they understand and accept the risk –
but still remain accountable for the
outcome.
4. How is the compliance requirement
verified? A privacy requirement, as an
example, may state that an organisation
only uses private data for the purposes for
which it was collected. This may translate
into a requirement that a cloud service
provider must contractually commit not to
use customer supplied data for any other
purpose than delivering the service.
Verification may be to confirm this is
included in the contract.
29
Privacy Compliance requirements related to privacy
may result from federal and state legislation,
industry regulations, codes of practice,
government policies and other sources.
Although this may lead to a great deal of
variety in the specifics, there are some
general principles that often surface. These
typically include restrictions on the
appropriate collection and use of private data,
the need to notify individuals about privacy
policy and seek their consent and the need to
effectively secure and maintain the integrity
of private data.
There may also be some controls on the
transmission of private data outside the
original jurisdiction. In these cases,
compliance requirements may require either
the explicit consent of subjects for their data
to be sent offshore, or a contract between the
organisation and the service provider that
ensures the privacy protections are
maintained by the service provider. The
compliance obligation generally rests with the
organisation that collected the data and does
not typically transfer to the service provider,
except as a set of contract terms they should
adhere to.
The following table highlights some of the
privacy related compliance requirements that
might exist and be relevant to a modern
application. Working through these questions
will help determine the compliance
requirements around privacy that need to be
factored in.
Question
Sources What acts of legislation, rules, regulations, policies or codes or practice exist that relate to how the organisation deals with private information?
Is there a Privacy Commissioner, ombudsman or other independent party that oversees compliance to privacy requirements?
Do the privacy compliance requirements make a distinction between private and sensitive data?
Current Practice
How does the organisation currently ensure their compliance with privacy requirements?
How will current privacy practices be impacted by deployment of the solution being evaluated?
Authority Who is currently accountable in the organisation for privacy compliance?
Specifics Will any personally-identifiable information be collected, processed or stored within the application?
Is employee information, collected, processed or stored?
Is any of the private information regarded as sensitive, private information, such as health records or financial data?
What principles and requirements govern the collection and use of private data?
What requirements exist around notice and consent?
Are there restrictions on the transfer of data out of the jurisdiction and if so, what are they?
What restrictions must be enforced on any service provider processing data?
What restrictions must be enforced on the device used to collect data, such as location and user details?
Are there any particular standards that need to be complied with, either for privacy impact assessment or privacy practices?
Verification How is compliance with privacy requirements demonstrated or verified?
Is there an external mechanism for auditing, certifying or attesting to privacy compliance obligations?
Figure 18: Privacy Compliance Questions
30
Records Many government and private sector
organisations have specific requirements to
maintain effective record-keeping systems.
Generally these requirements call for
important records to be protected from loss,
destruction or falsification.
Records generally should be categorised with
details of retention periods, formats, location
and media. Consideration needs to be given
to possible deterioration or loss and formal
procedures for handling, processing and
disposal need to be defined.
In government, record keeping requirements
are formally codified with additional
requirements often relating to freedom-of-
information requests, national security
classification and archiving.
In the context of modern applications, these
record keeping compliance requirements can
lead to concerns around:
how the integrity of information is
maintained within a cloud service through
mechanisms like backup and recovery
how information is classified, marked or
identified to ensure it can be located in the
future
how the service or data will be accessible
in the future for the duration of the
required retention period
how the integrity of the record keeping
system can be verified through audit or
other measures
how these audit mechanisms are
protected to ensure they cannot be
tampered with
how information held in particular formats
can be accessible in the future
how records be provided in the event of
legal discovery or law enforcement request
This table highlights some of the key
questions around records compliance that
should be worked through.
Question
Sources What acts of legislation, rules, regulations, policies or codes or practice exist that relate to how the organisation deals with records?
Is the organisation subject to freedom-of-information, archive disclosure or other government regulations?
Current Practice
How does the organisation currently ensure their compliance with record keeping requirements?
How will current record keeping practices be impacted by deployment of the solution being evaluated?
Authority Who is currently accountable in the organisation for record keeping compliance?
Specifics Are there formal information classification and marking schemes that the organisation complies with?
What requirements exist for the storage and retention of records?
Is the information that will be processed in the solution considered to be a formal organisation record?
Are there any restrictions on record formats or media?
How should disposal of data be handled?
What audit mechanisms need to be in place to ensure that records are not tampered with?
Is there a need for the organisation to deal with freedom-of-information requests or legal discovery on the information in the solution being considered?
Verification How is compliance with record keeping requirements demonstrated or verified?
Is there a mechanism for external auditing, certifying or attesting to record-keeping compliance obligations?
Figure 19: Records Compliance Questions
31
Information Assurance Many organisations have specific information
security assurance obligations that derive
either from government security policies,
sector-specific regulator policies or as a result
of the nature of their business operations.
Government agencies will typically have
centrally defined requirements such as the
Protective Security Policy Framework(PSPF) in
Australia or the Federal Information Security
Management Act in the US (FISMA). These
typically entail a range of security controls
that should be applied by an agency, a
mechanism to accredit a system’s security and
a management framework for accountability
and governance. Generally, the information
classification dictates how the extent to which
these controls are mandatory.
Private sector organisations that act as
suppliers to government might also need to
factor in government compliance obligations
into their thinking. Private sector
organisations that process credit cards, health
care information or other sensitive activities
may also have externally imposed information
assurance requirements, such as the Payment
Card Industry Data Security Standards
(PCIDSS).
In high security environments, particularly
those operated by national security
organisations of government and their
suppliers, explicit formal evaluation
procedures may apply. For example, software
or a device may need to undergo a formal
Common Criteria evaluation against a
specified Protection Profile in order for it to
be considered acceptable for use. Certain
devices, applications and cloud services may
not be permitted for use if these security
evaluations are not complete and current.
More commonly though, information security
assurance requirements derive from best
practices defined in standards like ISO 27001
and ISO 27002
Question
Sources What acts of legislation, rules, regulations, policies or codes or practice exist that relate to how the organisation secures information?
Current Practice
How does the organisation currently ensure their compliance with information assurance requirements?
How will current information assurance practices be impacted by deployment of the solution being evaluated?
Authority Who is currently accountable in the organisation for information security compliance?
Specifics What classification of data will be stored or processed within the solution being evaluated, and does this have a bearing on the level or extent of information assurance compliance obligations?
Do the information assurance requirements align to any existing international standard or best practice, such as ISO standards?
Are there formal processes for accreditation or certification that need to be adhered to?
To what extent are the requirements mandatory or optional?
What compliance requirements exist around physical security that need to be factored in?
What are the compliance requirements around personnel security, such as employee screening and clearance?
What are the compliance requirements around information security?
Are there specific requirements around how information security management is governed?
Do the information assurance requirements specifically make distinctions around outsourcing or offshoring of data?
Verification How is compliance with information assurance requirements demonstrated or verified?
Is there a mechanism for external auditing, certifying or attesting to these compliance obligations?
Figure 20: Information Assurance
Compliance Questions
32
Other Compliance Requirements These are just some examples of the types of
compliance requirements that exist. Only the
customer organisation can determine the
specific range of compliance obligations they
hold. Some other types of compliance
requirement might include:
Intellectual Property
Most countries have legal requirements
around copyright and other intellectual
property infringement. It’s important for an
organisation to understand precisely who
retains intellectual property ownership over
data stored in a cloud service and the possible
derivative uses of that data.
Procurement Processes
Many government agencies have specific
procurement practices such as standard terms
and conditions, panel contracts of approved
suppliers, authorisation limits and gateway
reviews. These processes may introduce
specific compliance requirements that need
to be factored into the assurance process.
Appropriate Use
Most organisations have compliance
requirements around the appropriate use of
information and communications technology.
In particular, services may be restricted from
being used for distribution of pornography,
sharing of child abuse images, publication of
terrorist material, etc. Similarly, organisation
may need to ensure their information and
communications technology is not used to
perpetrate fraud or other crimes, such as
sending out spam or malware.
Lawful encryption
There may be compliance restrictions on the
use of very high grade cryptographic
materials. Although these restrictions have
lessened and usually only affect national-
security related organisations, an organisation
may need to consider if these apply.
33
Security Architecture and Policy
Requirements Every organisation has existing technology
infrastructure and practices or intended
strategic directions that need to be
considered when evaluating the security
capabilities of a modern application. For
example, if an organisation has invested in
deploying a particular identity management
solution, it would be desirable for the cloud
and device solution being evaluated to be
able to integrate with this identity solution.
Note that these requirements are not the
business requirements of the specific solution
being considered, but are the security
requirements that would apply broadly to any
modern application to enable it to integrate
with the organisation. For example, this
would include acceptable mechanisms of
authentication and access control, required
support for encryption of data across the
network, mechanisms for identity integration,
necessary interfaces for audit and event
reporting and so on.
The three key areas of security architecture
and policy that need to be considered in the
evaluation are:
Identity & Access: How the organisation
registers new users, authenticates,
authorises and oversees access of users to
information
Management & Monitoring: How
infrastructure and applications within the
organisation are currently managed and
monitored
Information Protection: Specific
technologies or approaches used by the
organisation to protect information
The easiest way to identify these security
functional requirements is to work through
questions such as in the table below.
Identity & Access
Question
Authentication What forms of authentication are acceptable for the solution (eg. Passwords, certificates, one-time-passwords, hardware tokens, etc)?
Does the organisation currently use multi-factor authentication?
Does the device need to be authenticated as well as the user?
Will sensitive users such as administrators require higher assurance authentication?
Is single-sign-on a requirement for the solution through identity federation?
Are there specific technologies or standards required by the organisation for authentication, such as SAML, OAUTH, x.509, etc.?
Access Control
How does the organisation currently perform access control to applications?
What factors are incorporated in making an access decision, such as user identity, device identity, location, etc?
Is there a mechanism for centralised authorisation employed or intended to be used by the organisation?
Are there specific technologies or standards used by the organisation for access control, such as SAML, XACML, etc?
Provisioning & Lifecycle Management
How is the lifecycle management of identities within the organisation currently performed?
Is there a centralised technology or management system used by the organisation for identity lifecycle management?
Audit & Logging
How much data needs to be logged and audited within the solution? (for example, every login / every action / every attempted action, etc)
Are there existing technologies used by the organisation for logging or auditing?
Do the access logs need to be correlated with other sources of logs, for example physical access events with online access events?
Figure 21: Identity & Access Requirements
34
Management & Monitoring
Question
Configuration Management
What tools or systems does the organisation currently use for asset and configuration management?
How are configuration change processes currently implemented within the organisation?
Are there systems or applications within the organisation for which their configuration needs to be strictly controlled – for example, critical infrastructure components?
Security Management & Monitoring
What tools or systems does the organisation currently use for security incident and event management?
What existing incident response processes will need to integrate with the solution?
Will the organisation need to correlate events across cloud, on-premise and devices?
Service Management & Monitoring
Are there tools or systems that the organisation uses as standard for service management and monitoring?
How does the organisation currently track service availability and performance?
If an existing service became unavailable within the organisation, how would the organisation become aware and what processes would trigger?
Figure 22: Management & Monitoring
Requirements
Information Protection
Question
Encryption What requirements does the organisation have as standard for encryption of data-at-rest on devices or in the cloud?
What requirements does the organisation have for encryption of data-in-transit?
Are there specific encryption protocols, algorithms or key sizes required?
What key management practices are required?
Do the encryption keys need to remain controlled by the organisation? If so, does the organisation have appropriate key management and distribution infrastructure?
Threat Management
How does the organisation currently implement threat management (such as anti-malware, anti-spam, intrusion detection and prevention systems, etc)?
Will the organisation require traffic between clients and web services to traverse a controlled gateway?
If so, what restrictions are imposed at that gateway (ie. protocols, ports, white/black lists, etc)?
How does the organisation perform vulnerability scanning and address vulnerabilities discovered?
Data Loss Prevention
What are the organisations current practices around backup and recovery that are relevant to the solution?
Does the organisation currently implement or require data labelling or classification? If so, is this automatic or selected by the user?
Do security controls need to enforce policy such as data flow restriction, encryption or auditing based on labelling or classification?
How would the organisation currently deal with a data spillage event?
Does the organisation have any specific requirements for media sanitisation or destruction?
Figure 23: Information Protection
Requirements
Wo
rked
Exa
mp
le
35
Worked Example
External Requirements (Compliance)The following table of compliance requirements has been developed based on a review of key
documents:
1. NSW Cloud Services Policy & Guidelines, August 2013 (CSP&G)
2. NSW Digital Information Security Policy, November 2012 (DISP)
Unless otherwise noted, the authority for all compliance requirements is the Department’s Risk &
Audit Committee.
Source & Authority
Specifics Current Practice Verification
State Records Act 1998 and General Authority (GA35)
State Records are prohibited from leaving the state except under circumstances authorised by GA35, namely:
The Department must assess and address the risks involved in taking and sending records out of the State for storage or processing by a cloud service provider
The cloud service providers’ facilities conform to requirements in standards issued by the State Records Authority
Contractual arrangements and controls are in place for safe custody and preservation of records
Ownership of records remains with the Public Office
The arrangement should be monitored for conformance to standards.
The Department currently only processes and stores data within NSW on Department premises that conform to State Records standards
Internal audit
Possible external audit by State Records Authority
Privacy and Personal Information Protection Act 1998
Health Records and Information Privacy Act 2002
Agencies must take reasonable steps to ensure that no unauthorised access, use, modification or disclosure of personal information occurs.
The Department must ensure provisions for the following are in place:
The cloud provider must hold and process customer data consistent with the instructions of the Agency, and may not use the data for its own purposes.
Data security and safeguards against misuse, loss, unauthorised access or alteration must be in place
The Department and the data subject must be able to have ongoing access to the data. Access for the data subject will typically be via the Department.
Although the Department does not currently use cloud services, it does have processes in place to manage compliance with privacy requirements when dealing with third party contracted suppliers and service providers.
In these instances, the Department manages private information and third parties are prohibited from using or disclosing that information for any other purpose than contracted with the Department.
Full ownership of data remains with the Department.
All requests for access by a data subject to their private information are handled by the Department, not by the contracted supplier.
No formal verification requirements exist
Wo
rked
Exa
mp
le
36
Source & Authority
Specifics Current Practice Verification
Ownership of data must remain with the Public office on termination of the service
Data must be retained unless authorised to be disposed
Government Information Public Access Act 2009
Members of the public have an enforceable right to access government information. The specific relevant requirement is:
The Department must ensure that records and data created, stored and managed on the cloud remains accessible and retrievable. The cloud service provider does not respond directly to a request for information under this Act.
The Department has a formal process for responding to any freedom of information requests.
No formal verification requirements exist
NSW Digital Information Security Policy Memo
All NSW Government Departments must satisfy a set of information security management requirements:
Departments must have in place an Information Security Management System (ISMS) based on the application of ISO 31000 for risk management and ISO 27001 for information security management.
Agencies must comply with a specific minimal subset of controls across the domains of ISO 27001
All shared service providers within government must attain and maintain third party accreditation to ISO 27001
Each agency must attest annually to the adequacy of their information security
The Department is currently in the process of implementing improvements to the governance and practice of information security.
The minimal set of mandated controls are implemented but the Department has not obtained certification to ISO 27001.
Annual attestation for the Agency
Certification to ISO 27001 by third party independent accreditor
Procure IT Procurement Guidelines and Standards Terms
It is mandatory that all procurement processes, terms and conditions of contracting are consistent with the NSW Government Procure IT. This includes:
Processes for tendering, probity, evaluation and ordering of products and services that must be performed
Standard contract terms covering issues such as liability, indemnity, intellectual property, etc. that should be included in supply contracts
The Department currently leverages the ProcureIT framework. Where circumstances deviate, the Department can alter standard contract terms within appropriate authorisations.
Possible procurement review or audit
Wo
rked
Exa
mp
le
37
Internal Requirements (Security Architecture & Policy)The following table describes the internal security requirements of the Department as they relate to
each of the key areas of identity & access, management & monitoring and information protection.
The Authority for all requirements is the Department’s Chief Information Officer.
Identity & Access:
Source & Authority
Specifics Current Practice Verification
Internal Access Control Standards
Authentication:
For general users, username and password credentials are acceptable within the Department network
Two factor authentication using a smartcard or other factor is preferable for remote access to information
Two factor authentication is required for all administrator access to the service
The Department has a strategy to progress towards single-sign-on for all corporate applications. Cloud services will need to integrate by performing federated authentication
The Department favours the use of SAML 2.0
The Department currently only uses domain credentials or username/password combinations for authentication to applications.
Administrators must authenticate with a smartcard to work remotely, but internally can just use domain credentials.
Having single sign-on is desirable future state for the Department
None
Internal Access Control Standards
Access Control:
Access to applications must be authorised by an administrator based on a determination of business need
For the applications being assessed in this evaluation, all should be accessible remotely
Access to each corporate application is currently managed one by one by the corporate applications team.
The only application that is available remotely at present is email through a web access portal.
None
Internal Access Control Standards
Provisioning & Lifecycle Management:
Once a user is created in the corporate directory, they should automatically get access to the base level of applications, including email, collaboration and communications.
The Department should be able to manage groups of users with similar access rights, granting those groups access to applications as a group.
If an employee leaves the organisation, their account should be suspended internally which should then immediately revoke acess to cloud services.
Currently, each new user is provisioned within the Departments Human Resources system and we have a process for then requesting a system user to be configured in our corporate directory. Periodically, users who have left the organisation are removed from the directory.
None
Internal Security Audit Standards
Audit & Logging:
The Department requires every user login event to be logged and auditable.
The Department currently logs all login events and each application used within the Department has capabilities for application specific logging.
None
Wo
rked
Exa
mp
le
38
Source & Authority
Specifics Current Practice Verification
This at a minimum should include date/time and application accessed.
The Department requires all data deletion and modification actions to be logged and audited
It should be possible to selectively configure auditing for all read access
All logs should be accessible to the Department
Management & Monitoring:
Source & Authority
Specifics Current Practice Verification
Systems Management Standards
Configuration Management:
.The Department should be able to administer the configuration of the application and have advance notice of changes to the cloud service configuration that might affect users.
The Department has a configuration management process but no formal toolset that needs to be integrated with
None
Incident Response Standard
Security Management & Monitoring:
The Department should be able to configure and view reports of system access, malware detections or other security event.
The Department should be alerted to suspicious activity or attacks against Department applications
Currently, access to all external resources is monitored through a gateway. All malware incidents are reported though a security management application
None
Systems Management Standards
Service Management & Monitoring:
The Department must be able to access management reports on service availability and performance, tracked against committed service level agreements
Nominated system administrators in the Department should be alerted if any services become unavailable.
Service reporting information should be integrated with the Department’s existing service management toolset
The Department has in place a systems management tool for help desk, case management, alerting and executive reporting.
None
Wo
rked
Exa
mp
le
39
Information Protection:
Source & Authority
Specifics Current Practice Verification
Network Security Standard
Encryption:
Data should be encrypted at rest within the cloud service
It is desirable for there to be a dedicated network connection or encrypted virtual private network between users inside the corporate network and the cloud service
Users accessing the service from outside the corporate network should always have their communication encrypted using at least SSL3.0
It should be possible to selectively encrypt documents or communications based on the sensitivity or classification of the data within them.
The Department currently has limited use for encryption, except in the case of S/MIME for encryption of secure emails for one part of the organisation.
None
Incident Response Standard
Threat Management:
The Department requires all communications to be scanned for malware and spam prevention should be performed by the service provider
The service provider should be able to demonstrate intrusion detection and prevention capabilities within the cloud service.
Currently, the department has an antimalware solution installed on premise for email communications and using an external service for spam.
Intrusion detection is primarily performed at the network gateway where the Department has installed network firewalls and intrusion detection systems
None
Systems Management Standards
Data Loss Prevention:
The Department should be able to block or hold communications being sent to or from the cloud service. .
It should be possible to assign a classification to documents or other records using metadata
Currently the Department has little capability to detect or block flows of information out of the Department.
None
40
Step 3 – Verify Claims
Assurance Boundaries The intent behind defining assurance
boundaries is to break the solution down into
components about which a specific set of
security assurance claims can be made.
Usually, these are set by the limits of a party’s
span of control. For example, consider a
cloud service provider that operates
infrastructure-as-a-service to offer virtual
machines. The cloud service provider can
make claims about the security of the physical
data centre, the backup and recovery
mechanisms, security monitoring, the
virtualisation fabric, and so on. They may
conduct audits and obtain certifications like
ISO27001. But they are unable to make any
valid security claims about the applications
running within the virtual machines, about the
network connecting the virtual machines to
an end user or about the security of the end
user’s device.
A typical modern application that leverages
cloud services, devices, big data and social
networks may be composed of a number of
assurance boundaries. A realistic example is
illustrated below for a solution that provides
online learning and video conferencing. There
are at least eight assurance boundaries in this
solution:
1. Enterprise on premise infrastructure
including the devices and other systems
2. Enterprise cloud identity & collaboration
service which provides the services for
communication and sharing of information
3. Cloud application which provides the
learning application operated by a
different cloud service provider
4. Cloud application infrastructure which may
be operated by a third party cloud provider
5. Consumer identity service such as a
Facebook, Microsoft or Google identity
6. Consumer communications service such as
Skype enabling the consumer to perform
voice and video-conferencing
7. Consumer apps & devices in the home
through which they interact
8. App store which facilitates the discovery
and deployment of apps to the end user
Figure 24: Example of a complex solution with multiple assurance boundaries
41
Not all assurance boundaries need to be
assessed with the same level of rigour and
some can indeed be safely disregarded. In our
example scenario, most of the assurance
focus would be on assurance boundaries 2, 3
and 4 which compose the enterprise cloud
services for identity, collaboration and the
online learning application. It may however
be worth investigating the effectiveness of the
privacy practices within the consumer
communications service to ensure that the
consumer’s privacy is not adversely affected
(assurance boundary 5), or assess the integrity
of the consumer identity service (boundary 4)
or even the integrity of the app store
(boundary 7).
The intent is not to conduct an exhaustive
assurance process over every aspect of the
solution, but to map out the different spans of
control that exist over the security of different
parts of the solution. In the example solution,
every assurance boundary has a different
owner and only the owner can make
meaningful assurance claims. In this case,
the vendor of the online learning application
(ie. the owner of assurance boundary 3) can
make claims about the security of the
application but they also rely on the validity of
claims made by the cloud virtualisation
service on which the application runs (ie. the
owner of assurance boundary 4). However,
the service provider for the collaboration
service also operates the identity service and
can therefore make meaningful claims about
the security of both services together.
So the question then is what types of claims
can an assurance boundary owner make and
how do they relate to the functional security
requirements and compliance requirements
of the organisation.
Types of Assurance Claims Cloud service providers, application vendors
and device manufacturers seek to assure
prospective customers of the security of their
services using a variety of certifications, audits
and attestations. Examples include ISO 27001
certification, SSAE16 audit reports,
government certifications like FISMA and UK-
IL2 and compliance with privacy terms like
HIPAA, Safe Harbour, EU Model Contracts and
Data Protection Agreements or industry
assurance mechanisms such as those of the
Cloud Security Alliance and Open Data Centre
Alliance. Devices, applications and equipment
used within cloud services may also be
evaluated in formal schemes such as Common
Criteria or FIPS140.
Documentation provided by cloud service
providers in the form of technical
descriptions, security control frameworks,
contractual service descriptions, policy and
procedures documents can also be valuable.
Even though it may not be validated by a third
party, most service providers are careful to
only claim functions, capabilities and
processes that they do indeed perform.
Each of the forms of assurance claim have
varying intent, scope, applicability and
ultimately merit in verifying the security
assurance of a service, application or device.
There are some key considerations to keep in
mind when assessing the merits of any
particular claim:
Scope: The scope of an assurance claim may
be limited to certain facilities, locations,
functions, business units or other criteria. The
scope may not match the elements of the
solution within the assurance boundary.
Type: Some assurance claims relate only to
the management of security, others relate to
the actual observed practices. Some have
only a financial or operational focus (for
example SSAE 16 SOC1). Others are more
technical in nature such as FISMA
certification.
42
Authority: Service providers can be self-
audited or engage an external audit.
Generally, a reputable external auditor
provides a higher level of assurance. Other
claims require very specific certifying
authorities such as the US Federal
Government in the case of FISMA/FedRamp,
the UK Government in the case of UK-IL2, etc.
Applicability: Some assurance claims are
more appropriate to cloud services while
others relate to applications or devices. For
example, Common Criteria and FIPS 140 are
not relevant to cloud services as a whole but
be very relevant to devices and applications
used within a cloud infrastructure.
Validity: Some certifications are only valid for
a defined period of time and need to be re-
audited on a periodic basis. Others are only
valid for a particular configuration in
perpetuity but are no longer deemed valid if
the configuration changes.
The following is a high level guide to some of
the more common security claims.
Technical Documentation Service providers, device manufacturers and
application vendors commonly publish
technical information on their security
capabilities and practices. This may include
whitepapers, control frameworks and
implementation guides that provide
comprehensive security details. Most
reputable service providers take a substantial
amount of care to ensure that their
documentation is accurate.
As a starting point, this technical
documentation is probably the most useful
resource. The downside of technical
documentation is that it is unverified by an
external party so important security assurance
claims need to be verified by other means.
Cloud Security Alliance STAR Registry The Cloud Security Alliance (CSA) was formed
in 2008 as a not-for-profit organisation with a
mission to promote the use of best practices
for providing security assurance of cloud
computing services. It is comprised of quite a
broad coalition of industry practitioners,
corporations, associations and other key
stakeholders. As one of their first steps, the
alliance created the CSA Security, Trust &
Assurance Registry (STAR). This is a publicly
accessible registry that documents the
security controls provided by various cloud
computing offerings, thereby helping users
assess the security of cloud providers they
currently use or are considering contracting
with. Within the STAR registry, cloud service
providers publish details on how they
implement each of the controls within the
CSA Cloud Controls Matrix which is a
generalised security control framework for
cloud services.
There can be very significant variability in the
depth of information provided by different
vendors with some service providers simply
indicating compliance to controls, while
others provide extensive documentation on
how many controls are implemented.
The major drawback of the CSA Registry is
that all information is self-reported and non-
verified. To address this, the CSA has recently
introduced three new levels of reporting,
called CSA STAR Certification, CSA STAR
Attestation and CSA STAR Continuous.
Although few service providers have yet
adopted these additional levels, it is likely that
CSA STAR Certification in particular will gain
traction. This certification involves third party
audit against both the CSA Cloud Controls
Matrix and ISO 27001.
Further information on the Cloud Security
Alliance STAR is available at:
https://cloudsecurityalliance.org/star/
43
ISO 27001 Certification to ISO 27001 is perhaps the most
common assurance claim of cloud service
providers. Published in 2005, ISO 27001
formally describes an information security
management system with mandated
requirements across 11 different domains:
Security policy: the executive direction and
oversight around security
Organisation of information security: how
information security decisions are made
Asset management: inventory and
classification of information assets
Human resources security: employee
screening, training and exit
Physical and environmental security:
protection of the computer facilities
Communications and operations
management: technical controls around
exchange of information in networks
Access control: restriction of access rights to
networks, systems, applications and data
Information systems acquisition,
development and maintenance: how security
is built into the end-to-end lifecycle of
systems
Information security incident management:
preparing for and responding to information
security incidents
Business continuity management: ensuring
resilience of systems to disruption and
disasters
Compliance: satisfying local security policies,
standards, laws and regulations
A cloud service provider or other organisation
that has adopted ISO/IEC 27001 can be
formally audited and certified by an external
party as compliant to the standard. Upon
completion of the audit, the Cloud Service
Provider can provide a certificate of
compliance. The certificate will identify the
scope of certification which commonly
includes specific locations and services in
scope for the certification but generally
provides little additional detail. The Service
Provider should also be able to provide a
more detailed ISO 27001 Audit Report which
includes a summary of audit findings in each
of the 11 domains of the standard. This audit
report can be useful in determining if the
cloud service provider controls align to the
customer security controls.
Although the standard predates the
emergence of cloud computing and mobility,
the principles and requirements are
structured in a manner by which they are still
very appropriate. A revision of the standard
to ISO27001:2013 has recently been finalised
with changes that reflect an increasing focus
on risk management, security leadership and
executive focus on security outcomes. Some
controls have been updated to reflect modern
threats and security approaches. It will
however take a period of time for cloud
service providers and others to transition to
this revised standard.
44
SSAE 16 (SAS 70) – SOC & ISAE 3402 Statement on Standards for Attestation
Engagements no. 16 (SSAE 16) is a recent
standard put forth by the Auditing Standards
Board (ASB) of the American Institute of
Certified Public Accountants (AICPA). For
reporting periods ending on or after June 15,
2011, SSAE 16 became the new standard for
reporting on controls at service organizations,
essentially replacing Statement on Auditing
Standards no. 70, commonly known as SAS 70.
ISAE 3402 is a standard for reporting to which
SSAE16 complies. So to simplify – the
important assurance claim for a Customer to
focus on is SSAE 16, but there a few key points
to understand about how it works.
Firstly, SSAE16 provides for audit statements,
not certifications. So a service provider
cannot be ‘compliant’ to SSAE16 but rather
can provide an attestation to SSAE16.
Secondly, there are three different forms of
report under SSAE16. SOC 1 reports focus on
the internal control over financial reporting.
SOC 2 reports are more specifically designed
for reporting on the technical and procedural
controls of technology related entities,
including cloud service providers. Lastly,
similar to SOC 2 is SOC 3, which utilises a
defined list of Trust Services Principles to
report on cloud service provider practices in
five key areas:
Security: The system is protected, both
logically and physically, against
unauthorised access.
Availability: The system is available for
operation and use as agreed to.
Processing Integrity: System processing is
complete, accurate, timely, and
authorised.
Confidentiality: Information that is
designated “confidential” is protected as
committed or agreed.
Privacy: Personal information is collected,
used, retained, and disclosed in
conformity with the commitments in the
entity’s privacy notice and with the
privacy principles put forth by the
American Institute of Certified Public
Accountants (AICPA) and the Canadian
Institute of Chartered Accountants (CICA).
Thirdly, there are two ‘types’ of report: Type I
and Type II. Type I reports describe the
system and state that certain controls and
management practices are in place. Type II
reports build on this to indicate that these
controls and practices were consistently
applied and operating effectively over a
defined period of time.
Finally, the SSAE 16 standard requires a
written “assertion” by management. This
written assertion is essentially an assertion
made by the service organisation that must be
provided to the service auditor (i.e., the
auditing firm that is conducting the SSAE 16
engagement) representing and asserting to a
number of essential clauses, such as:
The description fairly presents the service
organisation's "system”
That the control objectives were suitably
designed (SSAE 16 Type I) and operating
effectively (SSAE 16 Type II) during the
dates and/or periods, as asserted to.
The criteria used for making these
assertions (which are additional
statements with supporting matter
regarding risk factors relating to control
objectives and underlying controls) were in
place (Type I) and were consistently
applied (Type II).
Customers assessing the security claims of a
particular service should request the SSAE16
SOC 2 or 3 Type II attestation if it is available.
This will be one of the most useful documents
in verifying the security claims of the service
provider. Similar to ISO 27001 certifications,
it is important to check that the attestation
scope is relevant to the systems within the
assurance boundary being considered.
45
FISMA / FedRAMP Authority to
Operate Based on experience with FISMA and the
lengthy time it takes for each agency to
evaluate and authorise operation of cloud
services for its own usage, the U.S. General
Services Administration launched an effort to
standardise security assessments of cloud
products and services across the government.
This program, called Federal Risk and
Authorisation Management Program
(FedRAMP), provides a government-wide
certification process to reduce costs and
duplication when multiple agencies attempt
to certify products and services for security
compliance.
In the past, each agency would typically take
applications and services through their own
FISMA accreditation process. FedRAMP
allows cloud services to be accredited once
and have that accreditation leveraged by all
federal agencies using what is called a
Provisional Authority to Operate (P-ATO).
The FedRAMP control set includes the NIST
800-53 R3 defined baseline for low and
moderate impact systems and a set of
controls that are FedRAMP requirements
above the baseline. These requirements have
parallels in the requirements of other
countries like the UK and Australia.
Federal government agencies in the US are
required to ensure that cloud service
providers they contract with have authority to
operate under FedRAMP. Further information
on FedRAMP and the list of cloud service
providers that have authority to operate can
be found at:
http://www.gsa.gov/portal/category/102371
US-EU Safe Harbour, Data
Processing Agreements and EU
Model Contracts There are a number of different but related
claims that a service provider may make in
relation to privacy.
The European Union/United States Safe
Harbor framework provides a means to help
U.S. organizations meet data transfer
requirements as needed by the European
Commission’s Directive on Data Protection.
Organisations self-certify that they implement
a comprehensive system of privacy
protections, permitting the transfer of private
information between EU and US data centres.
Moving beyond the Safe Harbour self-
certification mechanisms, in 2010, the
European Union released the EU Model
Contract Clauses (EUMC). These Model
Clauses help customers certify compliance
with the European Commission's Data
Protection Directive in order to allow them to
transfer data outside of the European
Economic Area. They include additional
security and notice requirements that a
service is willing to contractually commit to in
order to support customers. When included in
service agreements with data processors, the
Model Clauses assure customers that
appropriate steps have been taken to help
safeguard personal data, even if data is stored
in a cloud-based service centre located
outside the European Economic Area.
Some service providers will commit to data-
processing agreements (DPAs) for all
customers regardless of whether the
customer is located in the EU. This is because
broadly the EU data protection directives are
generally regarded as the high bar in privacy
protection globally. A customer organisation
that has particular sensitivity to privacy in
their assurance objectives should request the
service provider’s capability to contract to a
Data Processing Agreement and verify the
contract terms within that agreement.
46
Common Criteria & the Federal
Information Processing Standard Organisations that are familiar with
conventional approaches to security
assurance such as Common Criteria or FIPS
often look for a similar or equivalent
evaluation approach to cloud computing.
Existing government assurance requirements
may state a preference or need for
technologies used by government agencies to
be evaluated.
These models however have challenges in
being applied to cloud services and there is
currently no mechanism by which any cloud
service could be evaluated consistent with
Common Criteria. Technologies, devices or
applications used within cloud data centres or
network and computing equipment used to
access cloud services could however be
evaluated according to Common Criteria or
FIPS140.
Wo
rked
Exa
mp
le
47
Worked Example
External Requirements (Compliance)Building upon the external requirements defined in stage 2, the Department has undertaken a
review of technical documentation provided by cloud service providers, certification and
accreditation reports and contract terms. To assess option 3 (on-premise solution), the Department
assessed current practices in assessing our ability to meet compliance requirements.
The following table summarises the findings across the three alternative options.
Legend:
Fully conforms to requirement
Conforms with some additional contractual,
process or technical changes
Partial conformance to requirement
Non-conformant to requirements
Requirement 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
State Records are prohibited from leaving the state except under circumstances authorised by GA35, namely:
The Department must assess and address the risks involved in taking and sending records out of the State for storage or processing by a cloud service provider
Records are not leaving the state, so
risk assessment is not required
By reference to the risk assessment
section of this assurance evaluation, this requirement can be fully addressed
Stage 4 of this assurance process
(risk assessment) addresses this requirement
The cloud service providers’ facilities conform to requirements in standards issued by the State Records Authority
Current facilities on Department premises
conform to the standards requirements.
Technical description of local hosted cloud
provider facilities reviewed and found to be conformant to standards
Documentation review of physical security
measures provides sufficient assurance that physical facilities are conformant to requirements.
Contractual arrangements and controls are in place for safe custody and preservation of records
No contractual arrangement
necessary as records will be kept by the Department and preserved in an equivalent manner to current record keeping processes
Review of the standard contract
includes specific terms for the preservation of records
Review of the standard contract
includes specific terms for the preservation of records. Customer data held in the cloud can be destroyed 90-120 days after termination of the service, so the agency would have only that long to repatriate all data.
Ownership of records remains with the Public Office
No issue, as no data is leaving the
Department
Contract terms state that ownership rights
on all data remain with the customer.
For the public cloud provider being
evaluated, a standard contract term states that the Customer retains all right and ownership of the data.
The arrangement should be monitored for
Current monitoring practices are only
All data within the local hosted cloud can
Monitoring conformance to
Wo
rked
Exa
mp
le
48
Requirement 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
conformance to standards.
partially effective without complete visibility of information and records locations and classifications.
be effectively monitored with ease of access for physical inspection
standards will be achieved through annual certification to ISO 27001, but physical inspection or 1st party audit is not permitted
With regard to compliance of the Privacy of Personal Information Protection Act, the Department must ensure provisions for the following are in place:
The cloud provider must hold and process customer data consistent with the instructions of the Agency, and may not use the data for its own purposes.
No issue, as no data is leaving the
Department
Contract terms state that ownership rights
on all data remain with the customer.
Cloud service provider will sign a Data
Processing Agreement that explicitly states they will not use data for their own purposes
Data security and safeguards against misuse, loss, unauthorised access or alteration must be in place
The Department has work to do to improve
the current state of information security, particularly in relation to data loss prevention and authorisation of access
According to technical documentation
provided, the hosting provider does implement a comprehensive range of data security controls
Review of the technical
documentation, ISO 27001 audit report and SSAE16 SOC 2 attestations, the agency is confident the cloud service provider has an extensive range of data security controls in place.
The Department and the data subject must be able to have ongoing access to the data. Access for the data subject will typically be via the Department
No issue, as no data is leaving the
Department
The Department will be able to access
data within the service as required.
The Department will be able to access
data within the service as required.
Ownership of data must remain with the Public office on termination of the service
No issue, as no data is leaving the
Department
Contract terms state that ownership rights
on all data remain with the customer.
For the public cloud provider being
evaluated, a standard contract term states that the Customer retains all right and ownership of the data.
Data must be retained unless authorised to be disposed
No contractual arrangement
necessary as records will be kept by the Department and preserved in an equivalent manner to current record keeping processes
Review of the standard contract
includes specific terms for the preservation of records.
Review of the standard contract
indicates data will be retained unless authorised to be disposed, but can be destroyed 90-120 days after termination of the service if termination is by either party.
Members of the public have an enforceable right to access government information. The specific relevant requirement is:
Wo
rked
Exa
mp
le
49
Requirement 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
The Department must ensure that records and data created, stored and managed on the cloud remains accessible and retrievable. The cloud service provider does not respond directly to a request for information under this Act.
No issue, as no data is leaving the
Department
During the period of the contract, the
Agency has full access to data in order to respond to requests for information
During the period of the contract, the
Agency has full access to data in order to respond to requests for information
All NSW Government Departments must satisfy a set of information security management requirements:
Departments must have in place an Information Security Management System (ISMS) based on the application of ISO 31000 for risk management and ISO 27001 for information security management
The Department’s ISMS is not currently
based on ISO 27001 and there are a number of domains where the Department would have to invest to conform to this requirement.
According to technical documentation
provided, the ISMS of the cloud hosting provider is derived from ISO27001 and they use ISO31000 for security risk management
According to information security
policy and ISO27001 audit report provided, the ISMS of the cloud provider is derived from ISO27001 and they use ISO31000 for security risk management
Agencies must comply with a specific minimal subset of controls across the domains of ISO 27001
Currently the Department
implements a range of security controls consistent with the minimum baseline but some changes in identity & access management would be required to satisfy all requirements
According to technical documentation
provided, the hosting provider does implement the mandatory controls
The agency has verified through
review of the ISO27001 audit report that the cloud provider implements the mandatory controls.
All shared service providers within government must attain and maintain third party accreditation to ISO 27001
This is not presently an issue as the
service would not be considered a shared service
The hosted service data centre is certified
to ISO 27001, but the application for collaboration and communications is not.
The cloud service provider maintains a
contractual commitment to maintaining ISO27001 accreditation annually on the application and the data centre.
Each agency must attest annually to the adequacy of their information security
Currently the Department attests
based on information available without any formal audit or certification regime, but improvement in the security program is required to address gaps.
A review of the ISMS used by the hosted
service provider is adequate, but doesn’t have formal attestation according to SSAE16 or a similar standard.
The cloud service provider maintains a
contractual commitment to maintaining ISO27001 accreditation annually on the application and the data centre. In addition, we have reviewed the SSAE 16 SOC 2 reports of the cloud service provider which are a strong basis for attestation
It is mandatory that all procurement processes, terms and conditions of
Wo
rked
Exa
mp
le
50
Requirement 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
contracting are consistent with the NSW Government Procure IT. This includes:
Processes for tendering, probity, evaluation and ordering of products and services that must be performed
No issue, tendering for services to
implement the solution on premise can be conducted
The local hosted cloud provider is
familiar with our tendering processes.
Normal procurement processes for
tendering and evaluation present problems as the public provider is unlikely to respond to a market tender at fixed price, fixed spec.
Standard contract terms covering issues such as liability, indemnity, intellectual property, etc. that should be included in supply contracts
No issue with tendering for services
to implement the solution on premise can be conducted
Although the standard ProcureIT contract
terms need to be amended for a cloud service, the local cloud provider has worked with other government agencies locally on acceptable contracts.
Standard procurement contract terms and
conditions are not yet available in government and contract terms from the Cloud Provider are significantly different from ProcureIT
Internal Requirements (Security Policy and Architecture)The Department has reviewed the technical documentation provided by the alternative cloud
service providers. Workshops have also helped to assess the capability of each option to satisfy the
Departments’ needs.
The following table summarises the findings across the three alternative options.
Legend:
Fully conforms to requirement
Conforms with some additional contractual,
process or technical changes
Partial conformance to requirement
Non-conformant to requirements
Identity & Access
Requirement 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
Authentication
For general users, username and password credentials are acceptable within the Department network.
Current practice of using username and
password credentials can be extended to the new solution
Username and password
authentication supported
Username and password
authentication supported
Two factor authentication using a smartcard or other factor is preferable for remote access to information
The Department’s implementation of
two-factor auth is only for system administrators. An additional project would need
Based on technical document review,
two-factor authentication is possible to implement but it is an additional service.
Two factor authentication is
supported by the cloud service natively or it can be integrated with an on premise two factor authentication provider.
Wo
rked
Exa
mp
le
51
Requirement 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
to roll this capability out at substanial cost.
Two factor authentication is required for all administrator access to the service
Currently, this would require further
implementation work for the Department
Two factor authentication for
customer administrators is standard practice
Two factor authentication is
standard practice
The Department has a strategy to progress towards single-sign-on for all corporate applications. Cloud services will need to integrate by performing federated authentication
Although this strategy exists, it is currently
only at the planning stage
Standard practice of this hosting provider
The application includes an identity
governance capability
The Department favours the use of SAML 2.0
Currently none of the applications supported
by the Department use SAML 2.0, but an implementation of the new solution would involve SAML.
Based on technical document review,
applications with the hosted cloud either leverage SAML 2.0 already
SAML 2.0 is supported widely with
web-based applications on the public cloud
Access Control
Access to applications must be authorised by an administrator based on a determination of business need
Current in-house processes will
continue to apply
The Department will be able to control
provisioning of new users and granting access
The Department will be able to control
provisioning of new users and granting access
For the applications being assessed in this evaluation, all should be accessible remotely
There is currently minimal remote
access to applications from within the Department.
Applications will be accessible through a
browser or mobile device
Applications will be accessible through a
browser or mobile device
Provisioning & Lifecycle Management:
Once a user is created in the corporate directory, they should automatically get access to the base level of applications, including email, collaboration and communications.
The Department currently has no
mechanism to perform this automated provisioning and it is not feasible to include this within the project scope
A mechanism for automatic provisioning
of users within the cloud service with identity federation can be implemented within the scope
A mechanism for automatic provisioning
of users within the cloud service with identity federation can be implemented within the scope
The Department should be able to manage groups of users with similar access rights, granting those groups access to applications as a group
Currently access is granted on a
combination of groups and individual accounts
The Department will be able to control
access based on group membership
The Department will be able to control
access based on group membership
If an employee leaves the organisation, their account should be suspended
The Department does not currently have this
capability and it’s not feasible
Automatic de-provisioning of users
within the cloud service with
Automatic de-provisioning of users
within the cloud service with
Wo
rked
Exa
mp
le
52
Requirement 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
internally which should then immediately revoke access to cloud services.
to implement this sophistication of de-provisioning in the project scope.
identity federation can be implemented within the project scope
identity federation can be implemented within the project scope
Audit & Logging:
The Department requires every user login event to be logged and auditable. This at a minimum should include date/time and application accessed.
Standard capability The Department can have full access to
logs and work with the hosting provider to adapt logging
All login events are audited and available
for review
The Department requires all data deletion and modification actions to be logged and audited
Standard capability The Department can have full access to
logs and work with the hosting provider to adapt logging
Deletion and modification logs are
available but from a technical workshop, it is clear the logging capabilities are reduced
It should be possible to selectively configure auditing for all read access
The Department can reconfigure auditing
as required
The Department can have full access to
logs and work with the hosting provider to adapt logging
A standard set of logs are available.
Adapting those or additing additional log triggers is limited
All logs should be accessible to the Department
All logs are held on premise and so
always available to the Department
All application logs should be available
but network logs are unlikely to be provided in the event of an investigation
Only the standard set of logs would be
accessible to the Department.
Management & Monitoring
Requirement 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
Configuration Management:
The Department should be able to administer the configuration of the application and have advance notice of changes to the cloud service configuration that might affect users.
The Department can fully control the timing
of changes to the service
The Department would have advance
notice of changes but limited influence over their timing.
The Department generally may have
some advance notice of changes to the system functionality, but no ability to control the timing of those changes
Security Management & Monitoring:
The Department should be able to configure and view reports of system
The Department does not have a fully
cohesive security incident
The cloud provider reports on system
The cloud provider will report on system
access and malware
Wo
rked
Exa
mp
le
53
Requirement 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
access, malware detections or other security event.
and event management system.
access and malware detections
detections, but from a review of the technical specification and contract terms, it is understood the malicious attacks blocked by the cloud provider will not be reported.
The Department should be alerted to suspicious activity or attacks against Department applications
The Department does not have an
extensive detection mechanism
The Department generally will not be
notified of any suspicious activity or attacks unless they are successful.
The Department generally will not be
notified of any suspicious activity or attacks unless they are successful.
Service Management & Monitoring:
The Department must be able to access management reports on service availability and performance, tracked against committed service level agreements
The Department currently can monitor
and report on system availability
The hosted cloud service provider
provides online access to service availability and performance information
The Department has reviewed the service
availability and performance reports that are available
Nominated system administrators in the Department should be alerted if any services become unavailable.
Currently, the Department is able to
set up alerts for emergency situations
Administrator alerts can be set up based
on configurable events
Administrator alerts can be set up based
on configurable events
Service reporting information should be integrated with the Department’s existing service management toolset
It is expected the on-premise solution
would be be integrated with the existing tools
It’s possible to have integrated
management systems but significant configuration is required
It’s possible to have integrated
management systems but significant configuration is required
Information Protection
Requirement 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
Encryption
Data should be encrypted at rest if it is outside the corporate network
With an on-premise solution, the data is
being stored within the corporate network. However, the Department does not have a solution for the data being encrypted on remote or mobile devices
Email communications are
encrypted at rest, and all data in transit is encrpyted but other data may not be encrypted at rest
Email communications are
encrypted at rest, and all data in transit is encrpyted but other data may not be encrypted at rest
Wo
rked
Exa
mp
le
54
Requirement 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
It is desirable for there to be a dedicated network connection or encrypted virtual private network between users inside the corporate network and the cloud service
The service would be inside the corporate
network
A dedicated virtual private network can be
configured with the local hosted cloud provider, or a dedicated network at additional cost
A virtual private network can be
configured with the cloud service provider
Users accessing the service from outside the corporate network should always have their communication encrypted using at least SSL3.0
SSL/TLS used for all connections
SSL/TLS used for all connections
SSL/TLS used for all connections
It should be possible to selectively encrypt documents or communications based on the sensitivity or classification of the data within them.
The Department can incoorporate
classification marking into the solution on-premise but it’s not feasible to incorporate policy-based encryption
The hosted cloud provider has a
solution for marking of documents/emails as well as policy based encryption as an additional (added cost) feature
The cloud provider has limited ability to
support classification markings but can apply encryption by policy out-of-the-box.
Threat Management
The Department requires all communications to be scanned for malware and spam prevention should be performed by the service provider
The Department can continue to use a
combination of in-house and external antimalware and spam prevention tools
The local hosting provider solution
includes anti-malware and anti-spam
The cloud service provider incorporates
antimalware / anti-spam into the cloud service
The service provider should be able to demonstrate intrusion detection and prevention capabilities within the cloud service.
The Department has limited in-house
intrusion detection and prevention capabilities
The hosted cloud provider has outlined
their capabilities in intrusion detection and prevention, but with limited detail available it is difficult to ascertain the depth of capability available.
The cloud service provider has outlined
their capabilities in intrusion detection and prevention, but with limited detail available
Data Loss Prevention
The Department should be able to block or hold communications being sent to or from the cloud service.
Data loss prevention is not considered
feasible in the current scope
The hosted cloud provider supports this
type of data loss prevention but it requires significant configuration
The cloud service provider incorporates
data loss prevention out-of-the-box
It should be possible to assign a classification to documents or other records using metadata
The proposed solution will have the capability
to assign metadata classifications
Metadata classifications can be
applied to data stored in the cloud service
Metadata classifications can be
applied to data stored in the cloud service
55
Figure 25: Examples of risk events in each domain
56
Step 4 – Assess Risk
Domain Based Risk Assessment
In earlier stages of the assurance assessment,
the organisation develops a clear
understanding of the business goals, context
and assurance objectives. Then the
methodology goes on to determine specific
security and compliance features and verify
the cloud service provider’s security claims.
But modern applications that involve cloud
services, mobile devices, social networks and
big data externalise significant aspects of the
solution. Parts of the solution can be
operated in-house, other parts are run by
contracted service providers and other parties
in the supply chain of each service can play a
role.
This externalisation can bring risks that are
strategic, operational, compliance or technical
in nature. Strategic risks can impact the
organisation’s ability to fulfil it’s long term
goals or even lead to the demise of the
organisation. Operational risks can disrupt or
strain the functioning of the organisation
making it less effective or less efficient.
Compliance risks expose the organisation to
potential consequences from failing to satisfy
legal or regulatory requirements. Finally,
technical risks limit the technology choices of
the organisation or prevent it from achieving
best value from technology implementations.
The ‘domain based’ aspect of this risk
assessment is important. In order to
streamline the risk assessment, an approach is
taken to identify only the most important risks
in each of the assurance areas of
trustworthiness, resilience and adaptability.
For each of these three areas, risk events that
have strategic, operational, technical or
compliance impacts are catalogued and
assessed. This handbook details
approximately 50 of the most commonly
assessed risks but the organisation may
choose to add additional risks or determine
that some are not relevant to their situation.
The set of risk events encompass many of the
risks outlined in guides like the Cloud Security
Alliance Top Risks, ENISA Cloud Computing
Risk Assessment, Australian Signals
Directorate Cloud Computing Considerations
and other documents. They are spread across
each of the domain and impact areas as
shown in the diagram to the left.
Figure 26: Spread of risk events
across each domain
57
Leveraging ISO 31000
The risk assessment approach proposed in
this step is a process consistent with ISO
31000 to:
Understand the context: Revisit the
internal and external context developed
at step 1 in this framework
Identify risks: Involve a variety of
perspectives and stakeholders to
catalogue the relevant risks
Analyse risks: Assess the organisations
tolerance for a specific risk and determine
it’s likely probability and impact
Treat risks: Determine if there are risks
that require specific treatment and
consider possible mitigations
Provide an overall risk assessment:
Summarise the risk of each solution
option considered
This consistency with ISO 31000 enables
better integration with the organisation’s
broader risk management framework and
practices. It also brings along the consistent
terminology of risk tolerance, likelihood,
impact and exposure.
Risk Likelihood
Likelihood is simply the probability of a risk
event occurring on a scale from rare to almost
certain. Although every organisation can
develop their own rating scheme, a
reasonable guideline is provided below.
Risk Impact
Impact is the outcome or consequence of a
specific risk event on a scale from minimal to
catastrophic. Again, each organisation may
define their own rating scheme, but a useful
guideline is provided below.
Risk Exposure
Risk Exposure, sometimes referred to as the
risk rating, is simply the product of risk
likelihood and impact. In this framework, it is
rated on an extended scale from very low
exposure to extreme risk exposure. For
example using the diagram below, an
organisation that assessed a risk event as
having an unlikely probability, but major
impact would be rate their risk exposure as
significant.
Likelihood Illustrative Criteria
Almost Certain
‘Happens often’ or
‘Could occur within days or weeks’
Likely ‘Could easily happen’ or
‘Could occur in weeks or months’
Possible ‘Could happen, maybe has before’
‘Could occur in a year or so’
Unlikely ‘Never yet happened but might’
‘Could occur in 10 years or so’
Rare ‘Maybe in extreme circumstances’, or
‘A 100 year event’
Impact Illustrative Criteria
Catastrophic ‘Extremely negative coverage’ or
‘Unable to satisfy critical objectives’
Major ‘Persistent negative coverage’ or
‘Unable to achieve a core objective’
Moderate ‘Negative coverage for a few days’
‘Significant inefficiencies and loss’
Minor ‘Minor, transient negative coverage’
‘Inefficiencies somewhat recoverable’
Minimal ‘Isolated, brief coverage’, or
‘Slight inefficiencies’
Figure 27: Ranking scheme for risk likelihood and impact
58
Risk Tolerance
Risk tolerance is the concept of what level of
risk is the organisation is willing to accept to a
specific risk event. It is graded on an
equivalent scale to risk exposure.
In simplistic terms, if an organisation’s
exposure to a risk is greater than their
tolerance of that risk, then the risk needs to
be mitigated. If the risk is assessed to be
lower than the tolerance, then it is an
acceptable risk. In reality, the situation is not
so simple as the urgency with which a risk
must be addressed or it’s significance may be
a factor. For example, an extreme exposure
risk event for which the business has a low
tolerance may need to be immediately
addressed or it may even rule out any
possibility of a particular solution. Another
risk event for which the organisation has a
significant exposure but only a medium
tolerance may require less urgent attention or
scrutiny.
Comparing Options & Treating Risks
The focus on externalisation that comes with
assessing risk of cloud services can also draw
attention to existing risks that the
organisation is facing with current on premise
systems. Alternative service delivery models
may bring new risks while reducing others, so
the risk assessment needs to not only cover
the options involving cloud services, but also
have comparison to on premise or other
options. Apart from developing the
comparative risk assessment, a key objective
here is to identify risk events for which the
risk exposure is greater than the risk
tolerance. In these cases, the organisation
may investigate possibilities to accept, avoid,
reduce or transfer the risk.
Risk reduction strategies may include
implementing additional security mitigations,
including for example, encryption, access
controls, third-party monitoring or requesting
additional contract terms. Risk avoidance
decisions might include avoiding the transfer
of highly sensitive data to the service. Risk
transfer may include taking on insurance.
Ultimately, reasonable risk treatment
strategies should be incorporated into the
final analysis.
Figure 28: Risk Tolerance
Figure 29: Risk Tolerance and Treatment
59
Risk Event Catalogue
Strategic Risk Events
Risk Event Type Domain
S-Co-01: Cloud service data breach: A data breach of the cloud service provider results in sufficient data loss to cause strategic impact to the customer organisation
Strategic Confidentiality (Trustworthiness)
Risk Description:
If a major data loss occurs either broadly within the Cloud Service Provider, or specifically the Customer application/data hosted in the cloud service, then strategic impacts could include loss of stakeholder confidence, loss of market confidence, brand reputation damage, competitive disadvantage or financial loss.
Risk Tolerance Questions: Risk Exposure Questions:
How critical is the service and data to the strategic value of the business?
What strategic impact would a major breach of data stored in the service have on the organisation?
What if the breach became publicly disclosed?
What are the organisations current strategic risk considerations around data breach of existing systems?
What specific assurances can the Cloud Service Provider provide in terms of information security, such as certifications, practices, etc?
What is the scale and level of expertise available to the cloud service provider?
What would the consequences of a breach be for the Cloud Service Provider?
What evidence is there that they heavily invest in securing their systems?
What is their track record in preventing or dealing with any security breaches they might have had in the past?
What compensation would the Cloud Service Provider be liable for in the event of a data breach?
How does the Cloud Service Provider limit their liability?
Risk Event Type Domain
S-Tr-01: Cloud service bankruptcy: The service provider becomes financially non-viable or even bankrupt leading to degradation or termination of the service
Strategic Transparency (Trustworthiness)
Risk Description:
The service provider becomes financially non-viable or even bankrupt leading to degradation or termination of the service
Risk Tolerance Questions: Risk Exposure Questions:
How critical is the application or service to the operation of the organisation?
What is the current financial viability of the service provider and the depth of their financial resources?
Are their services profitable?
What is their credit rating and financial track record?
Do they have a proven track-record of investing and developing their services?
How large is their operation and what impact would a failure of their finances have on the market?
What is their track record of acquisition or mergers?
Are there any escrow mechanisms in the contract that deal with failure of the service provider?
60
Risk Event Type Domain
S-Tr-02: Inappropriate use of customer data The cloud service provider claims to protect privacy but uses customer data inappropriately or in ways that are inconsistent with the expectations of the Customer organisation
Strategic Transparency (Trustworthiness)
Risk Description:
Cloud services providers operate platforms on which data and works are created by Customer organisations, but contract terms of use may specify joint or service provider ownership of data and developments. If the Customer data is high value, this could create a strategic risk for the Customer organisation (Ref: ENISA R24, ASD CC 20i)
Risk Tolerance Questions: Risk Exposure Questions:
Will the service contain high value intellectual property?
Will the services developed by the Customer in the cloud be of high commercial value?
What are the organisation's current practices and expectations around intellectual property development, protection and commercialisation?
Does this intellectual property belong to the Customer organisation or are third parties involved?
What will the cloud service provider contractually commit to in terms of intellectual property ownership over data and developments?
Do all ownership rights remain with the Customer or are some transferred to the Service Provider?
What is the perceived value of those rights in the future?
Are there any relevant third party intellectual property claims?
On termination of the service, will all intellectual property stored within the service be destroyed?
Risk Event Type Domain
S-Pd-01: Termination by Service Provider Termination of service by the Service Provider due to acquisition or market circumstances
Strategic Predictability (Trustworthiness)
Risk Description:
Market failure, competitive pressures, funding challenges, acquisition, strategic portfolio changes or other factors may lead to service provider terminating a service or making significant changes so that it no longer satisfies the needs of the customer. (Ref: ENISA R5, R6)
Risk Tolerance Questions: Risk Exposure Questions:
How critical are the services to the organisation?
How quickly could the organisation adapt or find an alternative service if the service was terminated?
What is the scale and reputation of the service provider?
How established is the market for the service?
Are there competing standards or approaches that may not coexist well in the future?
Is the service expanding and being invested in?
What is the service providers track record around terminating services?
61
Risk Event Type Domain
S-Av-01: Extended period of service disruption The service fails and is unable to be restored to operation in a reasonable time causing strategic damage such as significant lost revenue, lost customers and brand damage
Strategic Availability (Resilience)
Risk Description:
If the service fails or is disrupted for a significant duration, the organisation may face strategic, long-term impacts that are likely to exceed caps on liability or other damages from the service provider.
Risk Tolerance Questions: Risk Exposure Questions:
How critical is the service to the continued operation of the business?
What impact would it have for the organisation if the service became unable for a prolonged period?
What are the business continuity and disaster recovery practices of the service provider?
What visibility can the Customer organisation have into these processes?
Will the Customer organisation maintain a separate backup or archive either on premise or within another Cloud Service Provider?
What options might be possible to return to operation with another service provider or on premise infrastructure if the cloud service provider could not return to operation?
Risk Event Type Domain
S-Fl-01: Service inflexibility to adapt to new needs Security or other requirements of the Customer organisation change, but the Cloud Service is inflexible to accommodate these changes
Strategic Flexibility (Adaptability)
Risk Description:
Changing external factors (such as new threats or opportunities) or internal factors (such as expansion, new ways of working, structure, etc) may emerge requiring the Service to be adapted. Some applications can be highly configurable initially, but inflexible to adapt. Such inflexibility may result in restrictions on the business or inability to effectively respond to changing circumstances
Risk Tolerance Questions: Risk Exposure Questions:
How dynamic is the organisation?
Does it typically need to respond rapidly to changing external or internal circumstances?
How will changes or reconfiguration of the cloud service be handled?
What is the cloud service provider’s track record around innovating and developing the service to deal with changing external factors, such as threats, authentication & access control options, etc?
Do changes or reconfiguration typically require downtime?
62
Risk Event Type Domain
S-Po-01: Non-portability of data on termination The service is terminated by either party, but the Customer is unable to extract the data in a way that can be ported to another service
Strategic Portability (Adaptability)
Risk Description:
The customer organisation decides to migrate to a different cloud application or service but is unable to readily extract the data in a format that is usable within the new service. Effectively they are locked-in to using the chosen application or service, not contractually but because the data itself is not portable. (Ref: ENISA R1, ASD CC 19m)
Risk Tolerance Questions: Risk Exposure Questions:
How likely is it that the organisation will switch to a different service?
If it occurred, how rapidly will that switch need to take place?
Are there multiple service providers in the market offering equivalent services?
Are there application interfaces and tools available for export and import of data?
Does the data export need to be performed by the service provider, the customer or a third party?
To what extent are open standards used for data structures?
Risk Event Type Domain
S-Cf-01: Unable to initially configure to fit The application or service cannot be configured initially to satisfy the business requirements of the Customer organisation
Strategic Configurability (Adaptability)
Risk Description:
Because cloud services work on the basis of configuring a common service, rather than customising to specific needs of a Customer, issues may emerge where either the service cannot be configured to requirement or the expense of configuring to requirement is too great.
Risk Tolerance Questions: Risk Exposure Questions:
How complex are the organisations requirements?
How well understood are they?
How adaptable is the organisation in changing business processes rather than the technology systems?
How configurable is the cloud service assessed to be?
Are there similar customers with equivalent scale, complexity, market and location already using the cloud service?
What is the Cloud Service Provider's track record of implementation?
63
Operational Risk Events
Risk Event Type Domain
O-Ex-01: Damage due to co-tenant activities Service disruption or loss of reputation due to co-tenant activities
Operational Exploitability (Trustworthiness)
Risk Description:
Resource sharing in the cloud service can mean that the activities of another tenant may affect the reputation of the customer organisation, for example due to spamming, fraud or serving of malicious content. Actions to mitigate or stop the malicious tenant activity may impact on the customer organisation.
Risk Tolerance Questions: Risk Exposure Questions:
How critical are the services to continuity of business operations?
What impact would it have if the organisation IP address range was shunned?
What is the scale and reputation of the service provider?
What are the policies and practices of the service provider for dealing with malicious tenants?
How does the service provider mitigate fraud or malicious hijacking of cloud services?
What activities are allowed / disallowed by the service provider – for example: do they allow vulnerability scanning or bulk emailing from their cloud service?
Are there any technical or operational mechanisms in place to prevent the activities of one tenant from impacting the reputation or service of another?
Risk Event Type Domain
O-Co-01: Malicious insider steals data Malicious insider within the cloud provider abusing high privilege access to customer data
Operational Confidentiality (Trustworthiness)
Risk Description:
Failure in internal cloud provider practices, ineffective screening, fraudulent intent or other motivation causing an internal administrator or other insider within the service provider to access and steal customer data. (Ref: ENISA R10, ASD 22b)
Risk Tolerance Questions: Risk Exposure Questions:
How sensitive is the data stored within the application?
Is there any sensitive or high value intellectual property stored within the service?
What would the impact be of a theft of customer data and disclosure publicly or to a competitor?
What practices does the cloud provider implement around employee screening and background checks?
How are these processes maintained during employment?
Are service provider employees security cleared, for example through a government security vetting program?
Do administrators have standing privileges to access customer data, or do those privileges need to be requested, authorised and then removed when no longer required?
What technical controls are in place to mitigate the risk of the abuse of high privilege roles?
Do third party suppliers have access to customer data or occupy high privilege roles?
What controls are in place to monitor third party supplier activities around use of higher privileges?
Are all activities of administrators logged, monitored and periodically audited?
64
Risk Event Type Domain
O-Av-01: Network disruption: The network between the cloud service provider and the Customer organisation endpoint is disrupted
Operational Availability (Resilience)
Risk Description:
With most cloud services relying on the internet for connectivity to customer endpoints, a broad or localised failure in network connectivity can temporarily disrupt the operations of the organisation. (Ref: ENISA R25)
Risk Tolerance Questions: Risk Exposure Questions:
How critical is the service to the continued operation of the business?
What would the impact be of a 1 hour disruption to operations? What about 1 day?
Is the service designed to be tolerant of temporary network failures, using mechanisms like caching, alternative network routes, client side processing, etc?
Does all network traffic need to pass through a common network gateway or are there multiple alternative network routes possible?
Are there multiple network routes between the data centre and the location where the customer is operating?
Can a user with the Customer organisation access the service even if the corporate network connectivity is offline, possibly using a public internet connection?
What service-level-agreements exist relating to the network interconnection between the cloud data center and the Customer organisation?
Risk Event Type Domain
O-Lt-01: Required capacity unavailable Service provider is unable to provide additional capacity on demand to meet customer need
Operational Load Tolerance (Resilience)
Risk Description:
Resource constraints in network, storage, processing or other areas may reduce the ability of the cloud service provider to provide the capacity requested by the customer. (Ref: ENISA R8, ASD CC 19l)
Risk Tolerance Questions: Risk Exposure Questions:
How variable is the resource load for the application being considered?
Is consumption of resources expected to peak far beyond normal levels under any particular circumstances?
In those circumstances, what would be the impact of disruption or performance degradation?
What is the scale of operations of the service provider?
Is the option being assessed involve on-premise or private cloud infrastructure where new hardware or resources would need to be acquired and commissioned?
Does the option utilise public cloud which typically has a much greater capacity to be provisioned on demand?
How long would it take to double the resource provisions of the service? How long to expand by a factor of 10?
Does the service provider have a track record of supporting applications at the scale required by the customer organisation?
65
Risk Event Type Domain
O-Lt-02: Network congestion: The network between the cloud service provider and the Customer organisation endpoint is congested or exhibits highly variable quality of service
Operational Load Tolerance (Resilience)
Risk Description:
With most cloud services relying on the internet for connectivity to customer endpoints, network congestion or poor quality of service can significantly impact productivity or end user experience within the Customer organisation. (Ref: ENISA R26)
Risk Tolerance Questions: Risk Exposure Questions:
How critical is the service to the continued operation of the business?
What impact would variable performance have on end user productivity?
Is the service designed to be tolerant of temporary network congestion, using mechanisms like caching, client side processing, etc?
What is the typical latency, congestion and packet loss characteristics of the network link between the Customer and data center locations?
Are there bottlenecks in the network topology that could cause congestion, such as a Customer Organisation-owned network gateway?
Are these gateways adequately specified for the transition of processing to an external cloud service?
What service-level-agreements exist relating to the network interconnection between the cloud data center and the Customer organisation?
Risk Event Type Domain
O-Tr-01: Confused security responsibilities Lack of clarity in security operational responsibilities leads to confusion and gaps in process that may ultimately lead to data loss.
Operational Transparency (Trustworthiness)
Risk Description:
Cloud service providers take on some, but not all of the security responsibilities for the application and the demarcation of responsibilities differs between across different service providers and types of cloud services. This can lead to gaps in process and accountability that may result in security events. (Ref: ENISA R20)
Risk Tolerance Questions: Risk Exposure Questions:
To what extent are the organisations current security practices consistent and standardised across the organisation?
How does the organisation currently deal with suppliers who may have different security practices and policies?
How well understood are the relative security responsibilities of the customer and the cloud service provider?
Are these responsibilities clearly documented and agreed?
How would a mismatch in expectations be resolved?
How experienced is the customer organisation in establishing outsourcing or external service provider arrangements?
66
Risk Event Type Domain
O-Ex-02: Unauthorised physical access: Access by an external attacker to premises, data or systems
Operational Exploitability (Trustworthiness)
Risk Description:
Unauthorised physical access by an attacker to machines, data or networking of the cloud provider could compromise Customer data leading to data loss, data integrity issues or disruption (Ref: ENISA R33, ASD CC 22d)
Risk Tolerance Questions: Risk Exposure Questions:
Tolerance questions: How sensitive is the data stored within the service?
What physical security controls exist within cloud service provider data centres, such as fencing, isolation, monitoring, guards, access controls, etc?
What certifications/attestations does the cloud service provider have relating to physical security?
What physical protections exist on the network infrastructure between data centres?
What physical cabling standards are enforced to help prevent accidental or deliberate incorrect cabelling?
Are third party facilities used by the cloud service provider, and if so, do the same level of physical protections apply to those?
How is data and equipment physically secured when outside of the data centre, for example during large data uploads / transfers or decommissioning of equipment?
What restrictions does the Cloud Service Provider place on third party visitation or external auditing?
Risk Event Type Domain
O-Co-02: Lawful confiscation of equipment Due to the shared resource nature of cloud services, if a lawful subpoena on another tenant leads to physical confiscation of hardware, then data of the customer organisation might also be disclosed
Operational Confidentiality (Trustworthiness)
Risk Description:
This is the scenario whereby if multiple tenants are on shared hardware and law enforcement require that hardware for evidentiary purposes in an investigation of one tenant, then the data of uninvolved tenants might also be confiscated and potentially be disclosed. (Ref: ENISA R21)
Risk Tolerance Questions: Risk Exposure Questions:
How sensitive is the information being stored in the cloud service?
What is the cloud service provider’s formal process for dealing with lawful requests?
Will they minimise the data or assets provided to law enforcement in a manner that excludes data of unrelated third parties?
Is the cloud service a public cloud (lower risk in this circumstance) or a private cloud (higher risk in this circumstance)?
Can dedicated resources be assigned to the customer?
In what jurisdictions are the resources located, and within those jurisdictions, how normalised are practices around lawful requests?
67
Risk Event Type Domain
O-In-01: Damage to logs Loss or compromise of operational or security logs
Operational Integrity (Trustworthiness)
Risk Description:
In the event of a security investigation or a legal request for information, complete logs are essential. If there is any tampering, if they are incomplete or if the cover an insufficient period of time, then the investigation could be hindered.
Risk Tolerance Questions: Risk Exposure Questions:
What is the sensitivity/classification of the data stored in the service?
How does the Cloud Service Provider implement backup and recovery?
What technical and process controls are put in place by the Cloud Service Provider to prevent compromise of the backup system or loss of backup media?
What is the retention schedule for backup media and how are they wiped at their end of life?
Risk Event Type Domain
O-Co-03: Accidental loss or theft of backups Backup media being lost
Operational Confidentiality (Trustworthiness)
Risk Description:
Because backups may contain all the data of the application, a compromised backup system or loss of backup media could lead to a large loss of Customer data
Risk Tolerance Questions: Risk Exposure Questions:
What is the sensitivity/classification of the data stored in the service?
How does the Cloud Service Provider implement backup and recovery?
What technical and process controls are put in place by the Cloud Service Provider to prevent compromise of the backup system or loss of backup media?
What is the retention schedule for backup media and how are they wiped at their end of life?
68
Risk Event Type Domain
O-Re-01: Disruption due to natural disaster A natural disaster such as an earthquake, fire, flood or extreme weather event disrupts the operations of the service
Operational Responsiveness (Resilience)
Risk Description:
If the cloud service provider or on-premise infrastructure is not able to rapidly respond to a natural disaster event and restore normal operations, then the Customer organisation may suffer downtime. (Ref: ENISA R35)
Risk Tolerance Questions: Risk Exposure Questions:
How sensitive is the data stored within the service?
What are the organisations current expectations and practices around disaster recovery and business continuity?
What is the likelihood of extreme weather events, earthquakes, fire, flood and other natural disasters in the region where the cloud service provider data centers are located?
What technical measures are in place to deal with disaster events, such as backup, geo-replication of data, redundant resources, alternative network routes, etc?
What is the scale and sophistication of the cloud service providers approach to business continuity and disaster recovery?
What is the cloud service provider's track record in handling disaster and outage events?
What assurance can the cloud service provider give that disaster recovery practices are effective?
Has the disaster recovery process been tested?
Risk Event Type Domain
O-Av-02: Downtime during business hours The cloud service is unavailable during periods of business operation
Operational Availability (Resilience)
Risk Description:
Lack of availability of the cloud service during business operations times may have a significant impact if prolonged or frequent. With regard to scheduled outages if required, the Customer organisation typically has less control over timing but may at least be notified in advance. (Ref: ASD 19f, 19g, 19h).
Risk Tolerance Questions: Risk Exposure Questions:
Tolerance questions: How critical is the service to the continued operation of the business?
What would the impact be of a 1 hour disruption to operations? What about 1 day?
What is the service level agreement guarantee provided by the cloud service provider?
What high availability measures are implemented by the Cloud Service Provider, such as clustering, replication, redundancy, etc?
How will the Customer organisation be compensated by the Service Provider if the service is unavailable beyond the SLA guaranteed time?
What limitations exist on the level of compensation provided by the service provider?
How is availability reported to the Customer?
What is the track record over the past year on service availability?
If there are scheduled outage time windows, how do they relate to the business day where the Customer is located?
69
Risk Event Type Domain
O-Lt-03: Variable performance under load The cloud service exhibits varying performance during periods of business operation due to resource contention within the cloud service
Operational Load Tolerance (Resilience)
Risk Description:
Lack of availability of the cloud service during business operations times may have a significant impact if prolonged or frequent. With regard to scheduled outages if required, the Customer organisation typically has less control over timing but may at least be notified in advance. (Ref: ASD 19f)
Risk Tolerance Questions: Risk Exposure Questions:
Tolerance questions: How critical is the service to the continued operation of the business?
What would the impact be of a 1 hour disruption to operations? What about 1 day?
What is the service level agreement guarantee provided by the cloud service provider?
What high availability measures are implemented by the Cloud Service Provider, such as clustering, replication, redundancy, etc?
How will the Customer organisation be compensated by the Service Provider if the service is unavailable beyond the SLA guaranteed time?
What limitations exist on the level of compensation provided by the service provider?
How is availability reported to the Customer?
What is the track record over the past year on service availability?
If there are scheduled outage time windows, how do they relate to the business day where the Customer is located?
Risk Event Type Domain
O-Cf-01: Lack of skilled resources Skilled resources are not available to configure or maintain the solution appropriate to the needs of the organisation
Operational Configurability (Adaptability)
Risk Description:
As with all information technology projects, the availability of skilled resources may present a risk. Although cloud services make it possible to draw on a wider pool of resources or offload work to a cloud service provider, skilled resource challenges may still exist.
Risk Tolerance Questions: Risk Exposure Questions:
How does the organisation currently deal with skills challenges?
Does the organisation have the budget and resources for skills development or recruitment?
Are skills from configuring and supporting existing applications transferrable to the new solution?
How prevalent are skills in the market for the technologies within the solution?
Does the cloud service provider have training, certification or skills readiness program?
Does the cloud service provider have consulting or engineering support staff locally?
70
Compliance Risk Events
Risk Event Type Domain
C-Co-01: Lawful warrant compelling disclosure Lawful warrant or subpoena process directed to the cloud service provider compelling the disclosure of customer data
Compliance Confidentiality (Trustworthiness)
Risk Description:
This is the scenario where a law enforcement or intelligence agency in the same jurisdiction as the Customer organisation, in the jurisdiction of the Cloud Service Provider data centre, or any other jurisdiction that applies extra-territorial powers makes a request of the Cloud Service Provider specifically for Customer data to be disclosed. As the cloud service provider would need to operate within the law of that jurisdiction, if it is a legally valid production order, Customer data could be disclosed. (Ref: ASD CC 20d)
Risk Tolerance Questions: Risk Exposure Questions:
What is the sensitivity / classification of the data being processed in the service?
What would the impact be of disclosing that data to a foreign law enforcement or intelligence service?
In what jurisdiction are the cloud service providers data centres? Where else could data flow?
How well regarded are the privacy protections and law enforcement/intelligence oversight processes in those jurisdictions?
What is the cloud service provider’s process for dealing with lawful requests for data?
What is their track record in complying with lawful requests for customer data?
Will the cloud service provider seek to direct requests to the Customer organisation if possible, rather than directly disclose information?
Can the customer restrict the allowable data location to specific jurisdictions?
Will the Customer organisation be notified if the jurisdictions in which data processing occurs changes?
Risk Event Type Domain
C-Co-02: Leakage of sensitive data into cloud Leakage of more sensitive data into the cloud service than it is designed/assured to contain
Compliance Confidentiality (Trustworthiness)
Risk Description:
This could be the scenario where although the cloud service is chosen and considered adequate to contain low or moderate business impact classified data, through user error, ineffective customer controls or usage creep, high business impact data is placed on the cloud. The most immediate impact may be that the organisation is unable to satisfy it's compliance obligations. (Ref: ASD CC 19i)
Risk Tolerance Questions: Risk Exposure Questions:
What is the sensitivity / classification of the data to be processed or stored in the service?
How does the organisation currently treat different classifications of data?
Are there different security assurance or compliance requirements depending on the classification of data?
How will the organisation segment or control data to ensure only appropriately classified data is processed in the cloud service?
Does the cloud service provider or the Customer organisation implement any technical controls for data classification or handling or data loss prevention?
In the event of a data spill (ie. higher classification data going to the cloud than is permitted), how would the customer detect the spill and what process would exist with the Cloud Service Provider to recover from and remove the spilled data?
71
Risk Event Type Domain
C-Tr-01: 3rd party supplier compliance issues Inadequate practices of a third party supplier to the service provider lead to customer organisation non-compliance
Compliance Transparency (Trustworthiness)
Risk Description:
The service provider engages a third party who doesn't adhere to equivalent policies or practices as the service provider, resulting in a significant compliance issue for the customer organisation. (Ref: ENISA R2, R7)
Risk Tolerance Questions: Risk Exposure Questions:
How highly regulated is the customer organisation?
Are there specific compliance areas relevant that are especially sensitive to third party or supply chain considerations, such as privacy?
How consistently are baseline practices currently monitored across suppliers?
Does the service provider engage third parties?
For what purposes are subcontractors engaged?
Does data transfer occur to subcontractors?
What level of data access would they have?
How assured are you that subcontracting agreements reflect terms in the customer contract?
Do you know who the subcontractors are and where they're located?
Will the service provider notify you of changing subcontractor arrangements?
Risk Event Type Domain
C-Tr-02: Failure to maintain adequate controls The service provider fails to maintain the security controls necessary for the customer organisation to satisfy required compliance obligations
Compliance Transparency (Trustworthiness)
Risk Description:
The customer adopts the cloud service with an understanding of certain controls in place that enable it to meet it's compliance obligations. However, the service provider either doesn't implement these controls or fails to maintain them adequately, leading to the customer becoming non-compliant. (Ref: ENISA R2, ASD CC 20c)
Risk Tolerance Questions: Risk Exposure Questions:
How highly regulated is the customer organisation?
Are there specific compliance areas relevant that are especially sensitive to third party or supply chain considerations, such as privacy?
Are there periodic or ad-hoc external audits of the customer security practices?
Does the service provider hold current SSAE16 SOC 2 reports that have been reviewed by the customer organisation?
Are there any relevant exceptions or limitations in those audit reports?
Is the scope of the audit reports appropriate for the solution being considered?
What are the service provider’s commitments around periodic audit and reporting?
What other certifications / audits does the service provider hold (eg. PCI DSS, ISO27001, etc)?
To what extent are the security controls described by the service provider explicitly articulated in formal service descriptions and contracts?
Will the customer be notified if a periodic external audit of the service provider identifies exceptions?
Does the customer have the ability to perform an audit / inspection?
72
Risk Event Type Domain
C-Pr-01: Insider accesses private data Malicious insider within the cloud provider abusing high privilege access to privacy-related customer data
Compliance Privacy (Trustworthiness)
Risk Description:
Specifically in relation to privacy-related data, such as personally-identifiable-information, financial information, credit history, health and social welfare information, a malicious insider could access and steal this data resulting a significant compliance issue for the customer organisation (ENISA R10, ASD CC 22b)
Risk Tolerance Questions: Risk Exposure Questions:
Is privacy-related customer data stored within the application?
Is this privacy-related data highly sensitive, such as financial details, health and social security data, etc?
What would the impact be of a theft of privacy related customer data and disclosure?
Is there a regulatory obligation for mandatory disclosure of privacy-related data loss?
What practices does the cloud provider implement around employee screening and background checks?
How are these processes maintained during employment?
What privacy training does the service provider organisation deliver internally?
Under what circumstances would a cloud provider administrator or other role be able to read privacy-related customer data?
Do any staff in the service provider have standing privileges to access privacy-related customer data, or do those privileges need to be requested, authorised and then removed when no longer required?
What technical controls are in place to mitigate the risk of the abuse of high privilege roles, specifically to access privacy-related data?
How are the privacy controls and expectations enforced across third party suppliers to the cloud service provider?
Is access to privacy-related customer data within the service provider organisation logged, monitored and periodically audited?
Risk Event Type Domain
C-Pd-01: Change in cloud provider practices: The service provider changes business practices or service functionality in a manner that no longer enables the organisation to meet compliance obligations
Compliance Predictability (Trustworthiness)
Risk Description:
Cloud services rapidly evolve but if the service provider makes significant changes, like for example changing their privacy policy to permit secondary use of data or terminating certain protections, then the organisation may no longer be able to satisfy compliance obligations.
Risk Tolerance Questions: Risk Exposure Questions:
Is the organisation operating in a highly regulated sector or have extensive regulatory requirements?
How dynamic is the organisation and capable of moving to a different service if necessary?
What is the service provider's track record in being predictable about the types of services being offered?
Have they significantly changed privacy or security policies in unpredictable ways?
What commitments can they provide contractually around maintaining certifications and minimum security & privacy protections?
73
Risk Event Type Domain
C-Pr-02: Inadequate privacy legal protections Cloud service provider may operate in a jurisdiction that provides inconsistent legal protections for privacy
Compliance Privacy (Trustworthiness)
Risk Description:
Cloud services operate on a global scale and in each jurisdiction must comply with local laws. The risk is that legal privacy protections or law enforcement powers are inconsistent with the customer organisation's jurisdiction to such an extent that the customer cannot be sure they are at all times satisfy their privacy protection obligations. (Ref: ENISA R22, ASD CC 19d)
Risk Tolerance Questions: Risk Exposure Questions:
Is private information or sensitive private information to be stored within the service? What are the organisation’s current compliance obligations in respect to privacy?
In what jurisdiction are the cloud service provider’s data centres? Where else could data flow?
How well regarded are the privacy protections and law enforcement/intelligence oversight processes in those jurisdictions?
What is the cloud service provider’s process for dealing with lawful requests for data?
What is their track record in complying with lawful requests for customer data?
What is contractually specified for data location in the service provider agreement?
Can the customer restrict the allowable data location to specific jurisdictions?
Will the Customer organisation be notified if the jurisdictions in which data processing occurs changes?
Risk Event Type Domain
C-In-01: Unable to respond to legal discovery Cloud service provider is unable to provide information required by the Customer Organisation to respond to a subpoena or warrant
Compliance Integrity (Trustworthiness)
Risk Description:
This is the scenario where a customer has to provide information in a legal matter or investigation, but due to limitations in the application, the Cloud Service Provider is unable to provide the required information in a timely manner.
Risk Tolerance Questions: Risk Exposure Questions:
What circumstances may lead to the customer needing to respond to a warrant or other legal request?
Is there information stored in the application that is considered likely to be subject to such a request?
What information can be provided by the Cloud Service Provider in the event of a legal request?
Does the service support legal hold, discovery or retention?
Will the Customer Organisation implement archiving on premise or with another cloud service provider?
74
Risk Event Type Domain
C-Pr-03: Cloud practices inadequate for privacy The cloud service provider claims to protect privacy but uses customer data inappropriately or in ways that are inconsistent with the expectations of the Customer organisation
Compliance Privacy (Trustworthiness)
Risk Description:
Cloud service providers potentially have access to large amounts of customer data that could be used in data mining, ad targeting, marketing or other purposes. The Customer organisation may not know this is occurring but if the private data of their clients is mined in this way, they may face a significant privacy compliance problem.
Risk Tolerance Questions: Risk Exposure Questions:
Will the service contain privacy related data, especially sensitive private data?
What obligations exist for the organisation to protect the privacy of their clients when engaging a third party for data processing?
What will the cloud service provider contractually commit in terms of use and explicit non-use of customer provided data (ie. commitment not to use Customer data for data mining, ad targeting, etc)?
Will the cloud provider commit to standard data-processing terms, such as the EU Model Contracts?
What formal certifications or assurances can the cloud service provider assert?
What is the reputation and scale of the cloud service provider?
Is private data encrypted or otherwise protected within the cloud service?
Risk Event Type Domain
C-Pp-01: Insufficient disaster preparedness Service provider business continuity preparedness is insufficient to satisfy regulatory requirements
Compliance Preparedness (Resilience)
Risk Description:
It is a common requirement of regulators or government that organisations have adequate preparedness for continued operations in the event of a disruption. If the cloud service provider can't demonstrate the continued adequacy of their preparedness, then the customer may face a compliance issue.
Risk Tolerance Questions: Risk Exposure Questions:
How critical is the service to the continued operation of the business?
What impact would it have for the organisation if a single user's data could not be restored when needed? What about for all data?
What are the business continuity and disaster recovery practices of the cloud service provider?
What visibility can the Customer organisation have into these processes?
Are they fully documented and are staff fully trained?
How frequently are these processes tested?
What audit reports are available from the service provider to demonstrate their preparedness?
How frequently will these audits be repeated?
75
Risk Event Type Domain
C-Fl-01: Unable to meet new compliance needs New compliance requirements emerge that cannot be satisfied by the functionality within the cloud service
Compliance Flexibility (Adaptability)
Risk Description:
New legislation, regulations, codes of conduct, supplier agreements and policies may emerge in the future that create new compliance requirements for the organisation. Alternatively, the Customer organisation may expand into new markets that impose compliance obligations. If the cloud service provider or the cloud service itself is unable to adapt to satisfy these new requirements, then the Customer may face a compliance problem or be limited from entering markets.
Risk Tolerance Questions: Risk Exposure Questions:
How sensitive is the data stored within the service?
What are the organisations current expectations and practices around disaster recovery and business continuity?
What is the likelihood of extreme weather events, earthquakes, fire, flood and other natural disasters in the region where the cloud service provider data centers are located?
What technical measures are in place to deal with disaster events, such as backup, geo-replication of data, redundant resources, alternative network routes, etc?
What is the scale and sophistication of the cloud service providers approach to business continuity and disaster recovery?
What is the cloud service provider's track record in handling disaster and outage events?
What assurance can the cloud service provider give that disaster recovery practices are effective?
Has the disaster recovery process been tested?
Risk Event Type Domain
C-Po-01: Contractual restriction on portability The customer is contractually limited from terminating the service and withdrawing their data or configuration for use in another service.
Compliance Portability (Adaptability)
Risk Description:
The customer decides to migrate to a different cloud application or service, but a contractual or legal restriction exists that prevents the extraction of data necessary to move to the new service. (Ref: ENISA R1, ASD CC 19l, 19m)
Risk Tolerance Questions: Risk Exposure Questions:
How likely is it that the organisation will switch to a different service?
If likely, how rapidly will that switch need to take place?
Are there terms in the contract that enable the service provider to assert ownership in any way over customer data?
Are there specific contract terms or termination provisions that would prevent the extraction of data?
76
Technical Risk Events
Risk Event Type Domain
T-Ex-01: Isolation failure between tenants Isolation failure enabling one tenant in the cloud service to exploit resources of another tenant
Technical Exploitability (Trustworthiness)
Risk Description:
Cloud services have multiple isolation mechanisms to separate storage, networking, memory and other resources of each tenant from one another. A failure in any of these isolation mechanisms may enable an attack leading to data loss or disruption. (Ref: ENISA R9, R19, ASD CC 21a)
Risk Tolerance Questions: Risk Exposure Questions:
How sensitive is the data stored within the application?
What would the impact be of a disruption to the service?
What is the track record of the cloud provider in implementing effective security practices?
Can they provide assurance around their secure development lifecycle approach?
What technical controls are in place to isolate storage, networking and memory?
How would the service provider detect and respond to an exploitation event?
What is the service providers approach for mitigating possible vulnerabilities in hypervisors?
How consistent and effective are the configuration, patch management and other infrastructure management capabilities of the service provider?
Are parts of the solution dedicated hardware or shared?
Are there any restrictions on which kinds of tenants can operate services within the cloud - for example a government community cloud?
Risk Event Type Domain
T-Co-02: Data visible on reallocated resources: Ineffective wiping of data resulting in possible data recovery by another tenant in the cloud service
Technical Confidentiality (Trustworthiness)
Risk Description:
If resources are scaled back, the customer terminates the service or for any other reason, resources are returned to a pool of available resources, there is a risk that data could still be held on the drive and visible to a subsequent tenant. This could lead to loss of customer data if a malicious third party attempted to recover inadequately wiped data. (Ref: ENISA R14, ASD CC20f, 21d)
Risk Tolerance Questions: Risk Exposure Questions:
How sensitive is the application/data to the organisation?
Does the cloud provider implement standard data wiping practices (such as NIST 800-88) to ensure that resources returned to a pool are appropriately wiped?
Is data encrypted-at-rest on drives?
77
Risk Event Type Domain
T-Co-01: Interception in-transit to customer Interception of data in-transit between customer endpoint and service provider
Technical Confidentiality (Trustworthiness)
Risk Description:
If data protection controls between the customer endpoint and the interface at the service provider are compromised, then data could be intercepted by a malicious third party.
Risk Tolerance Questions: Risk Exposure Questions:
How sensitive is the application/data to the organisation?
Are significant amounts of sensitive data transmitted between the customer endpoint and the service provider?
What encryption or other data protection measures are used to protect data in transit to/from the service provider? Specifically, is SSL/TLS or Virtual Private Networking used?
If on premise-servers are replicating or exchanging significant amounts of data with cloud servers, how is the data protected in transit?
Is information encrypted at the application or data level in addition to encryption at the transport level?
Are any counter-interception mechanisms used such as Perfect-Forward-Secrecy?
Is the cloud service provider able to provide assurance around how the service is protected against spoofing, replay, man-in-the-middle, sniffing and other attacks?
If large volumes of data are transferred out-of-band (such as mailing of hard drives for big data uploads), how are these transfers secured?
Can additional protections be implemented selectively for highly sensitive data, such as digital rights management?
What encryption protocols, algorithms and key lengths are used and are these consistent with requirements for information assurance compliance?
78
Risk Event Type Domain
T-Co-03: Interception between data centres Interception of customer data in-transit between service provider data centres
Technical Confidentiality (Trustworthiness)
Risk Description:
Replication and transmission of customer data between cloud service provider data centres is commonly performed to ensure high availability and disaster recovery. If data protection controls between one data centre and another are compromised, then data could be intercepted by a malicious third party. (Ref: ENISA R12,R13, ASD CC 20d)
Risk Tolerance Questions: Risk Exposure Questions:
How sensitive is the application/data to the organisation?
What encryption or other data protection measures are used to protect data in transit between service provider data centres?
Are these protections applied at the application level or on all network traffic?
Do the network connections cross international boundaries?
Can the customer specify that data transfer is restricted to certain regions or data centres?
Is data encrypted and unreadable even to the cloud service provider?
If the data is encrypted in transit between data centres, what are the key management practices of the cloud service provider?
What encryption protocols, algorithms and key lengths are used and are these consistent with requirements for information assurance compliance?
Risk Event Type Domain
T-Co-04: Interception due to loss of encryption Loss of encryption keys or weak encryption practices leading to interception or theft of data:
Technical Confidentiality (Trustworthiness)
Risk Description:
Encryption controls are only effective if keys are securely protected but a failure in key management practices could lead to loss of keys or weakening of encryption mechanisms (Ref: ENISA R17, ASD CC 22a)
Risk Tolerance Questions: Risk Exposure Questions:
How sensitive is the application/data to the organisation?
Are significant amounts of sensitive data transmitted between the customer endpoint and the service provider?
What are the key management practices of the cloud service provider?
If keys are retained by the customer to enable encryption/decryption on premise, how are these keys secured?
Are protocols like Perfect-Forward-Secrecy implemented to protect intercepted data even if keys are lost in the future?
What protocols, algorithms and key sizes are used for encryption at-rest and in-transit?
Which protocols are not permitted within the cloud service provider (eg, MD-5 for hashing)?
Are there circumstances by which a cloud provider would need to provide keys to law enforcement or other third parties?
79
Risk Event Type Domain
T-In-01: Backup failure The cloud service fails to create adequate backups or backed-up data cannot be restored when required
Technical Integrity (Trustworthiness)
Risk Description:
Backup data may be required to recover from a user error, system failure, security incident or legal investigation. If this data is not recoverable or only partially recoverable when needed, it may result in a data integrity problem for the Customer organisation. (Ref: ASD 19b,19c, 19j, 19k)
Risk Tolerance Questions: Risk Exposure Questions:
How critical is the service to the continued operation of the business?
What impact would it have for the organisation if a single user's data could not be restored when needed? What about for all data?
What are the backup and recovery practices of the cloud service provider?
What visibility can the Customer organisation have into these backup and recovery processes?
For how long are backups retained?
What would the process be for requesting the recovery of data from backups, in particular would it be performed by the Cloud Service Provider or customer administrators?
How long would it be expected to take?
How frequently and comprehensively are these backup and recovery processes tested?
Will the Customer organisation maintain a separate backup or archive either on-premise or within another Cloud Service Provider?
Risk Event Type Domain
T-Ex-02: Exploitation of customer gateway The communications gateway between the cloud service provider and the Customer organisation is exploited
Technical Exploitability (Trustworthiness)
Risk Description:
A cloud service solution may implement a gateway architecture, whereby there is some form of network protection and monitoring infrastructure at either the cloud provider boundary or the Customer organisation boundary. Compromise of either end of the gateway may lead to potential interception, traffic manipulation, disruption or creation of vulnerabilities that enable further exploitation. (Ref: ASD CC 19j, 19k)
Risk Tolerance Questions: Risk Exposure Questions:
Does the solution architecture incorporate a gateway implementation?
If the Customer organisation currently uses a gateway, how effective are the organisation's current mechanisms to protect the gateway itself from compromise?
Are there any specific certifications or compliance obligations existing for the organisation around the security of the gateway?
What technical controls are applied by the Cloud Service Provider in the gateway?
Specifically, do these controls include firewalls, content filters, antimalware scanners, intrusion detection, etc?
Are the gateway technologies accredited against relevant government or other security requirements?
Is it possible to establish a secure virtual private network with the Cloud Services Provider?
If so, how is that VPN secured, configured and monitored?
80
Risk Event Type Domain
T-Ex-03: Isolation failure on cloud fabric Isolation failure enabling one tenant in the cloud service to exploit resources of the cloud fabric
Technical Exploitability (Trustworthiness)
Risk Description:
Cloud services need to have isolation mechanisms in place to prevent a tenant developed application from breaking out of their sandboxed virtual environment to affect the infrastructure of the cloud service provider. A failure of this isolation mechanism that was sufficient to enable an attacker to gain access and modify the cloud fabric could lead to significant data loss or disruption. (Ref: ASD CC 21a)
Risk Tolerance Questions: Risk Exposure Questions:
How sensitive is the data stored within the application?
What would the impact be of a disruption to the service?
What is the track record of the cloud provider in implementing effective security practices?
Can they provide assurance around their secure development lifecycle approach?
What technical controls are in place to prevent virtualisation sandbox escape or mitigate other possible vulnerabilities in hypervisors?
How would the service provider detect and respond to an exploitation event on the service fabric?
How does the hypervisor solution protect resources of the host operating system from resources of the guest?
How consistent and effective are the configuration, patch management and other infrastructure management capabilities of the service provider?
Are there any restrictions on which kinds of tenants can operate services within the cloud - for example a government community cloud?
Risk Event Type Domain
T-Ex-04: Exploitation of cloud operations Exploitation of Cloud Service Provider administrative infrastructure leading to an attacker gaining access to Customer data
Technical Exploitability (Trustworthiness)
Risk Description:
Every cloud service provider operates an administrative infrastructure for monitoring, configuration management, helpdesk support, etc. An external attacker successfully compromising this administrative infrastructure could enable a compromise of Customer data (Ref: ASD CC 19g)
Risk Tolerance Questions: Risk Exposure Questions:
What is the sensitivity/classification of the data stored in the service?
What technical and process controls are put in place by the Cloud Service Provider to prevent exploitation of administrative infrastructure?
Specifically, are controls such as host configuration hardening, administrative privilege lock down, application whitelisting, patching, antimalware and other host protection controls apply?
What processes does the Cloud Service Provider have in place for detection and containment of any compromise on an administrative system?
Do administrators have standing privileges for access to cloud infrastructure and customer data?
81
Risk Event Type Domain
T-Ex-05: Customer infrastructure exploited Customer data held within the cloud application is lost due to exploitation of the Customer Organisation infrastructure
Technical Exploitability (Trustworthiness)
Risk Description:
A compromise of the infrastructure or user endpoints within the Customer Organisation due to inadequate security practices within the Customer Organisation could lead to loss of user or administrator credentials, and consequent disruption or loss of Customer data.
Risk Tolerance Questions: Risk Exposure Questions:
What is the sensitivity / classification of the data stored in the cloud service?
How assured is the Customer organisation currently of the adequacy of it's own security practices.
How will the endpoints and infrastructure within the Customer organisation that access the cloud service be protected?
How would a compromise of these systems be detected and contained?
What will the process be for management of privileged accounts within the Customer organisation?
Risk Event Type Domain
T-Su-01: Cascading failure in cloud An easily exploitable vulnerability within a component of the cloud service is discovered that can cause failure that replicates across the entire cloud service causing disruption or loss
Technical Survivability (Resilience)
Risk Description:
Cloud infrastructure at global scale is homogenously configured with the same operating systems, patch levels and applications. Hypothetically, a failure resulting from a discovered vulnerability could more easily cascade across a homogeneous system.
Risk Tolerance Questions: Risk Exposure Questions:
How critical is the service to the continued operation of the business?
What impact would it have for the organisation if the service became unable for a prolonged period?
What are the business continuity and disaster recovery practices of the service provider?
What visibility can the Customer organisation have into these processes?
What denial-of-service protections does the service provider implement?
How would the cloud service provider approach containing the spread of an attack across a set of homogeneous, internet-facing systems?
82
Risk Event Type Domain
T-Ip-01: Failed integration of security systems Existing or future security management and monitoring tools used by the Customer are incompatible with Cloud Service Provider systems
Technical Interoperability (Adaptability)
Risk Description:
Typically enterprise customers will need to maintain some level of visibility into the configuration and operations of the Cloud Service to integrate with an enterprise wide security incident and event management system, to correlate logs, maintain asset inventories, etc. The risk is that the Cloud Service Provider either doesn't expose sufficient data or exposes it in a way that can't be readily integrated into the enterprise's broader management toolset (Ref: ASD CC 19g)
Risk Tolerance Questions: Risk Exposure Questions:
What does the Customer organisation currently do in terms of security incident and event management?
Is there a standard toolset adopted by the Customer organisation for systems/security management and monitoring?
What impact would it have if those tools could not be integrated with the cloud service?
What security, management and monitoring data does the Cloud Service Provider expose for integration with Customer management systems?
Are there specific toolsets required?
Is the data provided in a non-proprietary manner that could be integrated with a variety of different management toolsets?
Are there third-party cloud service providers that can provide management & monitoring?
83
Threat Modelling
The risk assessment described so far is based
on an approach that looks across the widest
practical range of risk domains and possible
impacts. It’s intent is focused on breadth of
coverage, rather than depth.
However, some risk events may require a
more extensive examination that looks
deeper into the possible conditions that may
realise a risk and one that more fully
examines the security mitigations in place to
reduce the event’s probability or impact. This
may particularly be the case for risk events
with a potential major or catastrophic impact
to the organisation. For these high impact
risks, the organisation needs to be very
confident that their assessment of probability
is correct.
Additionally, for some risks to be realised
there may need to be multiple failures in
controls. For example, the risk of interception
of data transmitted between two cloud data
centres may require (a) failure of physical
controls on communication networks, (b)
failure of encryption methods on the same
communication networks. Determining the
probability of such multiple event scenarios
can be complex but necessary to inform an
assessment of that specific risk. This is where
threat modelling, and in particular the
concept of a threat tree can be very valuable.
A threat tree (sometimes referred to as an
attack tree) is a conceptual hierarchy showing
how an asset can be attacked, or more
formally how a risk event can be realised.
First described by Edward Amoroso and then
popularised by Bruce Schneier, threat trees
have gained wide adoption in areas where the
interrelationships between complex risk
events need to be assessed.
A threat tree has a single root node that
represents a risk event, and then multiple
child nodes that represent conditions or
events that must be satisfied for the root risk
event to be realised. Each child node can in
turn have multiple child nodes. The
relationship between a node and it’s child
nodes can be expressed logically as AND or
OR. That is, a node’s conditions can be
satisfied if a number of child nodes are true
(AND) or if one of it’s conditions are true (OR).
This is best represented by the simplified
example below. In this diagram the root risk
event is for readable customer data to be
compromised. For this to occur, firstly the
data would need to be compromised but it
would also need to be readable. If the data is
not encrypted, the root risk event could occur
if there was either an accidental data spillage,
deliberate compromise or lawful data
request. However, if the data were
encrypted, then not only must the data be
compromised, either the encryption keys
must become known to the attacker or the
encryption is found to be weakly
implemented.
Figure 30: An Example Threat Tree
84
Security controls or mitigations can be applied
at any level of the threat tree to reduce the
probability of a condition being satisfied. It is
indeed possible to map security controls
against each leaf of the threat tree in order to
assess their effectiveness in addressing the
root risk event. Following on from the
previous example, the diagram below maps
some illustrative process and technical
security controls to the threat tree.
Assessing the probability of any of these
conditions being satisfied (and therefore the
probability of the root risk event occurring)
then comes down to assessing the
effectiveness of each of these controls. This
approach can also be very effective in
understanding the benefit of adding a specific
mitigating control or having additional
controls applied by the cloud service
consumer. For example, the diagram shows
the inclusion of data loss prevention filters
which would be configured by the service
consumer rather than the cloud provider.
Every organisation will have different
perspectives on key risk events, the conditions
they see as precursors to the risk event and a
framework of technical and process controls.
As an aid, the handbook provides threat trees
and control mappings for three of the most
commonly assessed complex, high impact risk
events:
Readable customer data compromised
disclosed to a third party
Permanent customer data loss or damage
Disruption of service
It should be reiterated that because of the
depth of understanding required, this type of
threat modelling should be applied selectively
to high impact or complex risk events only.
85
Threat Tree 1: Overview
The first threat tree provides an overview of
the three key threats modelled and how they
are composed into sub-trees. A number of
threat trees are expanded in further diagrams,
so this only shows the root risk events and
their highest level children. Interconnections
between threat trees are not shown here but
they do exist and are described in the
subsequent more detailed models.
The diagram to one side and the following
tables describe each threat tree and aligned
controls.
86
87
88
Threat Tree 2: Customer Data is
Readable
The second threat tree in this hierarchy
focuses on the conditions by which data could
be in a readable, unencrypted state. There is
a common understanding that encryption is a
desirable control to protect data stored in a
cloud service. The assumption is that even if
data is lost or intercepted, it is meaningless
while in an encrypted state. However, for this
assumption to be valid, the encryption /
decryption keys must not become known, the
choice of cryptographic controls must be
adequate and cryptographic policies need to
be technically enforced appropriate to
different data classifications. These are the
conditions assessed in this threat tree.
The first condition for customer data to be
readable is simply that the data is not
encrypted in the first place. Encryption at rest
and/or in transit are commonly applied
controls. Local data encryption is a variant of
this practice, whereby data is encrypted on
premise before it is stored in the cloud and
decrypted again on premise prior to access by
a user. The challenge with local encryption is
that it makes it impossible for the cloud
service to perform any function more than
simply storage. Effective data classification
and an enforced policy on the use of
encryption for appropriate classifications of
data is also necessary.
Encryption keys could be lost if the cloud
provider infrastructure is compromised by an
external attacker (under circumstances
modelled in threat tree 4) and there is a
breakdown in key management practices.
This is because standard key management
practices isolate keys as highly sensitive
information including storage of the keys so
they cannot be exported by an attacker.
A similar circumstance could occur if
encryption keys are held by the customer but
the customer infrastructure is compromised.
It is common thinking that encryption is a very
effective control in cloud services, provided
the customer can retain control of the keys.
In reality, this depends on what threat the
customer is more concerned about and the
relative security capabilities of the cloud
service and the customer. If the most salient
threat is the possibility of a malicious act by
the cloud provider or one of it’s employees,
then the logic may be sound. However, if the
most significant threat is of a targeted attack
by a determined intruder, then considering
the relatively stronger security controls and
processes of cloud service providers in
comparison to their customers, this
assumption is likely to be untrue. It’s
probably far easier for an intruder to steal
keys from the customer organisation than to
steal them from the cloud service provider.
Finally, encryption may be rendered useless if
poor choices are made in the protocol,
algorithms and key lengths selected. For
most organisations, this becomes equivalent
to ensuring that modern systems are used
and kept up to date, but for more sensitive or
classified data, specific protocols, algorithms
and key lengths might be specified.
89
Identifier Event / Condition Aligned Controls
1.2 Customer data is readable
1.2.1 Data is not encrypted
Encryption at Rest: Encrypting data at rest can protect against loss of media or physical attack but it should be noted that the data is generally decrypted to be processed or transmitted by an application in the cloud, so encryption at rest is not effective against attacks on the application. Encryption in Transit: Encrypting data as it traverses a network between services and consuming devices or between data centres can be an effective control against interception. Local Data Encryption (Customer): Encrypting data before it enters the cloud service can be performed by a client application, encryption gateway or device. The benefit is that with all data encrypted in the cloud, the risk of data loss is reduced. However, this imposes severe restrictions on the utility of cloud applications: As they cannot access the data, the function of the cloud is limited to merely storage. Data Classification: Effective data classification, and then the enforcement of policy based on that classification is a fundamental control relevant to the use of the cloud. For example, a policy might enforce the rule that all data classified as Sensitive must be encrypted-at-rest in the cloud service. Generally, the classification needs to be applied by the Customer of the service, not the cloud service provider. Policy on Use of Cryptographic Controls: Based on data classifications, a policy on the use of cryptographic controls should describe how cryptography is used to protect data at different classifications.
1.2.2 Encryption keys become known
1.2.2.1 Encryption keys lost by cloud provider
1.2.2.1.1 (AND) Cloud provider infrastructure compromised
Step to Threat Tree 4: 1.1.3.1
1.2.2.1.2 (AND) Breakdown in key management enabling attacker access to keys
Key Management: Encryption controls are only as effective as their associated key management practices. A formal process needs to be in place covering the lifecycle of keys including key generation, replacement, revocation, access controls, storage and monitoring. Protected Key Storage: An important part of the overall key management practice is stored in a protected manner. The sensitivity of keys requires additional controls including isolation, access control and monitoring. Some scenarios also permit storage of keys in the cloud using a High Security Module, whereby keys are accessible to the application, but remain under the control of the customer.
1.2.2.2 Encryption keys lost by the customer
1.2.2.2.1 (AND) Customer infrastructure compromised
Step to Threat Tree 5: 1.1.3.2
1.2.2.2.2 (AND) Breakdown in key management enabling attacker access to keys
Key Management (Customer): As per the control described above, if the customer retains control of keys, then they must similarly apply a comprehensive key management practice.
90
Identifier Event / Condition Aligned Controls
Protected Key Storage (Customer): As per the control described in 1.2.2.1.2 above, but in this case the customer is maintaining the keys and therefore must take similar steps to ensure they are protected.
1.2.3 Data is weakly encrypted
Cryptographic Standards: Cryptographic protocols and algorithms differ in their ability to protect information and poor choice of algorithms, protocols, key lengths and other characteristics can render encryption controls to be of limited value. Therefore, it’s necessary to set specific baselines for each form of encryption and ensure they are applied.
91
92
Threat Tree 3: Accidental Data
Spillage & Law Enforcement Request
An accidental data spillage or release of
information to law enforcement as a result of
a warrant are two circumstances by which
customer data might be disclosed to a third
party. Limiting both of these circumstances
relies heavily upon having in place effective
procedural controls.
In the case of accidental spillage, the threat
model focuses on circumstance whereby data
might physically leave a secure area on media,
where an operator may accidently spill data
during the course of their work, where an
application itself may inadvertently spill data.
One other scenario for data spillage
considered is leakage if one tenant reuses a
resource previously used by another tenant,
such as a disk, and is able to see data of the
previous tenant. A range of process and
technical controls are described to address
these circumstances including the use of
dedicated systems, resource isolation, media
wiping, access management and encryption.
Disclosure due to a law enforcement request
is a very different scenario. In this
circumstance, the customer is the subject of a
warrant or other order that the cloud service
provider responds to. Clarity around
applicable legislation and formal business
processes to review, challenge and respond
are the most appropriate controls in this
regard.
93
Identifier Event / Condition Aligned Controls
1.1 Customer data disclosed to a third party
1.1.1 Accidental data spillage
1.1.1.1 Data on media leaving cloud provider secure area to uncontrolled site
Media Wiping: Wiping of media is necessary to eliminate all traces of data on a disk prior to it leaving a secure area. Wiping typically involves multiple successive overwrites of zero or random data on to the disk. Encryption at Rest: Encryption of data stored on media is an effective control against spillage of data.
1.1.1.2 Unintended spillage directly from cloud application
Secure Development Lifecycle Practices: Threat modelling and other secure design techniques can help ensure that threats to software applications are identified and mitigated. This can include gaps in input validation that are a common cause of application data spillage.
1.1.1.3 Data on media, memory or network shared resources spilling
Resource Isolation: Cloud service providers routinely share resources such as networking, memory and storage with tenants isolated from one another using virtualisation approaches. Media Wiping: Wiping of media before it is reallocated to another tenant. Wiping typically involves multiple successive overwrites of zero or random data on to the disk. Encryption at Rest: Encryption of data stored on shared media Dedicated Systems: Some cloud services can be provided in a dedicated manner so that tenants do not have to share the same physical memory or storage.
1.1.1.4 Accidental spillage by cloud support personnel during operations
Data Loss Prevention: Monitoring outbound communications or network traffic to identify data spillages or deliberate attempts to exfiltrate data can reduce the risk of spillage Privileged Access Management: Privileged accounts by their nature have the ability to perform administrative activities to delete, remove or modify data and change configurations to disrupt the service. It’s therefore essential to have formally enforced mechanisms to control the allocation and use of these privileges. The most effective approach is to have zero standing privileges, only permitting an administrator to be granted a privilege on request for a specific purpose and duration. Operations Policies & Restrictions: Cloud services have service management, administrative and support teams with access to the cloud infrastructure. Technical protections on the systems they use such as authentication, hardening, antimalware, etc and process controls such as training, clearance, etc can be effective in reducing the risk of an accidental or deliberate data spillage.
1.1.2 Data disclosed to law enforcement
1.1.2.1 (AND) Customer is the subject of an order, warrant or other request
Jurisdiction for Data: The location of data stored in the cloud may have bearing on the ability of law enforcement to request access. However, in practice, multi-lateral assistance treaties may render this to be of limited relevance.
94
Identifier Event / Condition Aligned Controls
Customer Business Practices (Customer): The most significant factor in considering whether an organisation is likely to be subject of a law enforcement request is the governance, appropriateness and legality of their own business practices.
1.1.2.2 (AND) Cloud service provider responds to the warrant by releasing customer data
Legal Review, Challenge and Response Practices: The cloud service provider should have formal processes to review any request from law enforcement, determine it’s validity, and if necessary provide the minimum data requested. It may also be appropriate for the cloud service provider to redirect the request to the Customer or challenge the request.
95
96
Threat Tree 4: Cloud Service
Provider Infrastructure
Compromised
This is a complex portion of the threat tree
examining the set of conditions that could
lead to data disclosure through a compromise
of the cloud service provider infrastructure.
There are four major sets of circumstances
that are considered.
Firstly, the actions of a malicious employee or
contractor who deliberately seeks out and
removes customer data. Mitigating controls
like employee screening, monitoring,
privileged access management and data loss
prevention approaches can be effective in this
circumstance.
Secondly, the threat tree breaks down some
of the circumstances that may permit a
network exploitation by a remote attacker,
including attacks from one tenant to another,
an attack on the cloud fabric itself, an attack
on the operator systems or an attack on the
Customer application running in the cloud. A
range of relevant controls related to isolation,
secure development practices, vulnerability
management and intrusion detection &
mitigation are listed.
Thirdly, the threat tree considers the
circumstances by which an attacker could gain
access by using valid credentials that are
either lost or stolen or perhaps attack the
authentication mechanism itself, for example
by brute force guessing of passwords. Control
emphasis here is on credential management
processes and technically stronger
authentication mechanisms like multi-factor
authentication.
Finally, physical attack is considered whereby
an attacker would need to breach physical
protections to access systems, locate
customer data and extract that data. Control
emphasis here is on a range of physical
protections for the data centre itself as well as
processes to securely dispose of or handle
data if it goes off-site.
97
Identifier Event / Condition Aligned Controls
1.1.3.1 Cloud service provider infrastructure compromised
1.1.3.1.1 Malicious employee or contractor within cloud provider extracts data
1.1.3.1.1.1 (AND)
Malicious employee or contractor has sufficient access rights
Privileged Access Management: Privileged accounts by their nature have the ability to perform administrative activities to delete, remove or modify data and change configurations to disrupt the service. It’s therefore essential to have formally enforced mechanisms to control the allocation and use of these privileges. The most effective approach is to have zero standing privileges, only permitting an administrator to be granted a privilege on request for a specific purpose and duration. Administrator Monitoring & Audit: The actions of administrators should be logged, monitored and periodically audited. Employment Screening: Background verification, employment history, skills and qualifications checking are all valid ways to attempt to identify employees with potential malicious intent. Segregation of Duties: Risks associated with a user-role conflict of interest, such as the same person being able to request a privilege and be able to authorise the granting of that privilege can be addressed through careful segregation of duties.
1.1.3.1.1.2 (AND)
Malicious employee or contractor has the ability to remove data
Data Loss Prevention: Monitoring outbound communications or network traffic to identify attempts to exfiltrate data. This can also be applied to removable media. Operations Policies & Restrictions: Cloud services have service management, administrative and support teams with access to the cloud infrastructure. Technical protections on the systems they use such as authentication, hardening, antimalware, and process controls such as procedures to formally request access rights to data can be effective.
1.1.3.1.2 Network exploitation by external attacker to gain access to data
1.1.3.1.2.1 Access gained via a web application attack
1.1.3.1.2.1.1 Successful attack on cloud application’s management interface or API
Secure Development Lifecycle Practices: Threat modelling and other secure design techniques can help ensure that threats to software applications are identified and mitigated. This can include vulnerabilities in the handling or validation of data by application interfaces (API’s)
1.1.3.1.2.1.2 Successful attack on cloud application
Secure Development Lifecycle Practices (Customer): If the application is developed by the consumer of the cloud service, as in an IaaS or PaaS based application, then the customer needs to take due care in the development of the application.
1.1.3.1.2.2 Attack on cloud infrastructure resulting in access privileges
1.1.3.1.2.2.1 Tenant-to-tenant attack resulting in access privileges for attacker
Tenant Isolation: All cloud services make extensive use of virtualisation approaches in order to share physical resources like memory, storage and processing power across multiple tenants.
98
Identifier Event / Condition Aligned Controls
Isolation of these tenants from one another is critical to preventing interference from one tenant to another. Network Segmentation: Segmenting traffic on networks can mitigate the risk of attack from a tenant on one network segment against a tenant on another segment. Typically in cloud infrastructures, this segmentation is achieved virtually rather than physically. Domain Isolation: Domain isolation, sometimes referred to as community cloud, is about restricting the tenants on the cloud infrastructure to be from a specific community, such as a government cloud. This approach typically has limited effectiveness and it depends heavily on processes to restrict entrants to the community cloud. Intrusion Detection & Breach Mitigation: Processes and tools to detect network and processing anomalies, investigate and correlate security events, contain incidents and continuously improve security approaches are essential. In this scenario, it is important that attempts to attack another tenant from within one tenant are detected and contained.
1.1.3.1.2.2.2 Attack on tenant resulting in access privileges for the attacker
1.1.3.1.2.2.2.1 Unpatched vulnerability in the tenant operating system
Virtual Machine Patch Management (Customer): If the solution is Infrastructure-as-a-Service (IaaS), then the cloud customer needs to ensure that patches are deployed and have a mechanism to do so. PaaS and SaaS patch management is performed by the cloud solution provider. Vulnerability Management (Customer): This is a process of proactively identifying vulnerabilities on a continuous basis, then assessing appropriate mitigations, such as patches.
1.1.3.1.2.2.2.2 Unpatched vulnerability in the tenant application
Secure Development Lifecycle Practices (Customer): Threat modelling and other secure design techniques can help ensure that threats to software applications are identified and mitigated. If the application is developed by the customer (as in IaaS or PaaS), then this control is implemented by the cloud customer. If the solution is a SaaS service, then the control should be implemented by the cloud service provider.
1.1.3.1.2.2.3 Attack on cloud fabric resulting in access privileges for the attacker
1.1.3.1.2.2.3.1 Unpatched vulnerability in the cloud fabric operating system
Cloud Fabric Patch Management: Monitoring the cloud underlying infrastructure and deploying appropriate patches Vulnerability Management: In relation to patches, vulnerability management comprise the processes for assessing and determining the need to deploy patches. Intrusion Detection & Breach Mitigation: Processes and tools to detect network and processing anomalies, investigate and correlate security events, contain incidents and continuously improve security approaches are essential. In this scenario, it is important that
99
Identifier Event / Condition Aligned Controls
attempts to attack another tenant from within one tenant are detected and contained.
1.1.3.1.2.2.3.2 Unpatched vulnerability in the cloud fabric management application
Secure Development Lifecycle Practices: Each cloud platform requires underlying infrastructure and software to manage virtual resources, commonly known as the cloud fabric. Threat modelling and other secure design techniques can help ensure that threats to the cloud fabric are identified and mitigated.
1.1.3.1.2.2.4 Attack on operator systems enabling privileged access to data
Privileged Access Management: Privileged accounts by their nature have the ability to perform administrative activities to delete, remove or modify data and change configurations to disrupt the service. It’s therefore essential to have formally enforced mechanisms to control the allocation and use of these privileges. Network Segmentation of Management Systems: Segmenting management traffic from normal traffic on the network can limit the potential for leakage of data via the management interfaces of systems. Operations Policies & Restrictions: Cloud services have service management, administrative and support teams with access to the cloud infrastructure. Technical protections on the systems they use such as authentication, hardening, antimalware, etc and process controls such as training, clearance, etc can be effective in reducing the risk of an accidental or deliberate data spillage. Intrusion Detection & Breach Mitigation: Processes and tools to detect network and processing anomalies, investigate and correlate security events, contain incidents and continuously improve security approaches are essential. In this scenario, it is important that attempts to attack operator systems and networks are detected and contained.
1.1.3.1.3 Use of valid credentials by external attacker to gain access to data
1.1.3.1.3.1 Valid cloud service provider administrator credentials lost or stolen
Multi-factor Authentication for Cloud Administrators: The use of multiple factors, such as username/password along with biometrics or a smartcard can be effective to reduce the risk of lost or stolen credentials. The most common factors used by cloud providers are biometrics for physical access and smartcards for system access. Credential Management: This comprises the process controls around issuance, revocation and protection of credentials.
1.1.3.1.3.2 Customer user or administrator credentials lost or stolen
Multi-factor Authentication for Customer Users (Customer): Customers of cloud services can also use multiple factors for authentication such as smart cards, tokens, etc. Credential Management (Customer): Processes need to be established to control the issuance, protection and revocation of credentials used by administrators and staff of the cloud customer.
1.1.3.1.3.3 Authentication mechanisms are vulnerable to attack
Authentication Policies: Authentication policies define the appropriate credentials required for access to different services. For example, low sensitivity assets may only require user name and password to access. Highly sensitive systems may require multiple factors for authentication. Also some authentication approaches are
100
Identifier Event / Condition Aligned Controls
more susceptible to attack such as for example, passwords with a small number of characters.
1.1.3.1.4 Physical attack by external attacker to gain access to data
1.1.3.1.4.1 (AND)
External attacker can gain physical proximity to access and remove data
Secure Perimeters: Physical perimeters protected with fences, doors, locks, barriers and guards are a foundational control for data centres to prevent damage or theft of equipment. Secure Areas: Typically data centres comprise of segmented areas, with different levels of sensitivity and access restriction. Physical Security Monitoring: Monitoring of physical security perimeters and internal areas with alarms, intruder detection systems and guards. Secure Disposal & Off-site Handling: If data or equipment is moved off site for backup, maintenance or other purposes, it is important to ensure that the data is protected. This may require additional measures such as encryption, wiping or policies that prevent off-site handling. Secure Equipment Management: The selection, commissioning and maintenance of equipment needs to performed in a manner that protects the security of data stored on that equipment
1.1.3.1.4.2 (AND)
External attacker can identify the specific physical location of data
Data Sharding: Data sharding is a process of distributing data across multiple locations in small portions rather than placing all data on a single drive.
101
102
Threat Tree 5: Customer or 3rd Party
Infrastructure Compromised
The extensive investment made by cloud
service providers in their security technologies
and practices makes them a hardened target
for any attacker. To a very real extent, it be
much wiser for an attacker to focus on a
potentially more vulnerable means of attack,
the customer infrastructure itself.
Compromising customer systems, apart from
gaining access to information still stored on
the customer on-premise systems, could
enable an attacker to steal credentials for
access to the cloud service, steal data
synchronised between the cloud service and
customer devices, or modify configuration on
customer infrastructure to disrupt access to
the service.
There are a wide range of mechanisms by
which an attacker could compromise
customer infrastructure from seemingly
simple approaches involving social
engineering to more sophisticated techniques
using custom malware. It may also involve
physical attack or perhaps the accidental or
malicious actions of an employee. It is
beyond the scope of this handbook to address
all these possibilities and relevant controls.
Instead, the threat tree focuses on how a
compromise of customer infrastructure could
be escalated to affect the customer services
or data contained within the cloud service.
There are two key scenarios considered here.
The first scenario is one in which data is
synchronised locally and accessible to the
attacker. The second highlights the risk that
an attacker who has stolen credentials on the
customer infrastructure could use those
credentials to access the cloud service.
It is important to note there are other
connections between this scenario of
customer infrastructure compromise and
other threat trees. For example, if an attacker
can gain access to customer held encryption
keys, then some of the encryption practices in
the cloud can be nullified.
Another area covered in this threat tree is
that of compromise of a third party or other
systems not under the control of either the
customer or the cloud service provider. This
could for example include the public or
private network between the customer and
the cloud provider or the network between
cloud service provider data centres. It may
also include third party services that integrate
with the solution, such as for example a third
party identity service provider or third party
suppliers that are embedded within the
supply chain, such as hardware equipment
manufacturers. Encryption, along with virtual
private or dedicated private networks are key
controls to consider to mitigate risk of these
scenarios.
103
Identifier Event / Condition Aligned Controls
1.1.3 Deliberate compromise leading to disclosure of customer data
1.1.3.1 Cloud service provider infrastructure compromised
Step to Threat Tree 4: 1.1.3.1
1.1.3.2 Customer infrastructure compromised
1.1.3.2.1 (AND) Customer infrastructure is compromised by an intruder
Customer Infrastructure Technical Controls (Customer): A range of customer technical controls covering governance of information security, security awareness, access control, data protection, physical security and so on are necessary to protect the customer infrastructure from compromise. Ideally, a comprehensive framework such as ISO 27001 is adopted. Customer Infrastructure Process Controls (Customer): To complement technical controls a range of management and procedural controls around personnel, physical and technical security are required on the customer side.
1.1.3.2.2 (AND) Intruder on customer infrastructure gains access to cloud data
1.1.3.2.2.1 Data is copied locally and accessible to the attacker
Data Loss Prevention: Monitoring outbound communications or network traffic to identify data spillages or deliberate attempts to exfiltrate data can reduce the risk of spillage.
1.1.3.2.2.2 Attacker gains credentials that enable access to cloud services
Multi-factor Authentication for Customer Users (Customer): The use of multiple factors, such as username/password along with biometrics or a smartcard can be effective to reduce the risk of lost or stolen credentials. An attacker might be able to get access to credentials like a username and password but if they also need a device like a phone or smartcard, then they are prevented from gaining access. Credential Management (Customer): This comprises the process controls around issuance, revocation and protection of credentials.
1.1.3.3 Compromise of 3rd Party systems outside cloud provider or customer control
1.1.3.3.1 Compromise of cloud service provider supply chain
Supply Chain Integrity Processes: Generally these are equipment and service acquisition controls to ensure that equipment or services required by the cloud service provider are evaluated as not introducing security risks.
1.1.3.3.2 Interception of communications between data centres
Encryption in Transit: Encrypting data as it traverses a network between services and consuming devices or between data centres can be an effective control against interception. Private Networks: The bandwidth of communications between data centres though may make encryption of all traffic cost-prohibitive. Private networks may be a reasonable alternative, although a preferable approach is to use both private networks and encryption. Interception Detection & Mitigation: Processes for monitoring, detecting and mitigating attempts to infiltrate or intercept communications between data centres.
104
Identifier Event / Condition Aligned Controls
Restrictions on Data Location (Customer): The customer may be able to choose in which locations data may be stored, such as restricting data to the local country or region. Although not directly a control that may impact the risk of interception, some may view different global regions as having a greater or lesser risk of interception.
1.1.3.3.3 Interception of communications between cloud customer and data centre
Encryption in Transit: Encryption of data between the cloud service provider and the customer is a foundational control necessary for any cloud service. Commonly this is implemented as SSL at the application layer Virtual Private Networks: A virtual private network can isolate and protect communication between the customer end point and the cloud service. Typically this is established through the use of IPSec, which is a standard for encryption and authentication between the end points at the network layer. Dedicated Network Connections: Some cloud services will be able to provide direct, dedicated network connections so that communications between the cloud service and the customer do not pass over the pubic internet.
105
106
Threat Tree 6: Permanent Loss of
Customer Data
The loss of customer data through deletion,
damage or destruction is a second risk event
that is commonly necessary to analyse.
Compliance requirements around record
integrity and maintenance as well as the
potential business impact of losing important
data generally drive the focus on this risk.
According to this threat tree, there are really
three conditions that need to simultaneously
exist for the risk event to occur.
Firstly, customer data has to be deleted,
damaged or destroyed. This could happen as
a result of an accidental deletion or damage
by the customer for which there isn’t a simple
‘undo’ process or feature. Alternatively, data
could be accidentally deleted or damaged by
cloud service personnel. The practice of
continuous replication of data across multiple
locations can mean that any deletion or
damage introduced can rapidly replicate to all
other locations where that data is stored. To
avoid this circumstance, change control and
privileged access management controls are
vital on both the cloud service provider and
customer side. Another alternative would be
the deliberate action of an attacker as
described in threat tree 4.
The second prerequisite would be for there to
be no readily recoverable backup on the
customer or 3rd party location. Some
customer may deem risks to be so significant
as to require replication of data on their
premise or to a 3rd party location. If such a
backup exists, there may still be disruption
and complexity to recovering the backup but
at least data is not irrevocably lost.
The third condition is for no readily
recoverable backup to exist within the cloud
service. Generally, cloud service providers
undertake to have comprehensive backup and
recovery controls in place including data
replication between paired sites, but there
are some circumstances in which no backup in
the cloud could be recoverable.
Firstly, if there was a simultaneous disaster in
the two paired sites of a cloud service, then
both production and failover data could be
destroyed. Although the prospect may be
remote, this is one reason why both sites
should be geographically separated.
Alternatively a disaster in just one site
combined with a failure of disaster recovery
processes, perhaps because those processes
were untested, could also lead to the risk
event occurring.
A final scenario whereby no readily
recoverable backup exists in the cloud would
be where backup and recovery processes are
inadequately implemented or insufficiently
tested. Customers may believe that backups
are being taken but they may not be
operational.
107
Identifier Event / Condition Aligned Controls
2 Permanent loss of customer data
2.1 (AND) Customer data is deleted or destroyed
2.1.1 Accidental deletion by customer
Customer Change Control Processes (Customer): Formalised processes for the planning, authorisation, execution and verification of changes in a controlled manner
2.1.2 Accidental deletion by cloud service personnel
Cloud Provider Change Control Processes: Formalised processes for the planning, authorisation, execution and verification of changes in a controlled manner Privileged Access Management: Privileged accounts by their nature have the ability to perform administrative activities to delete, remove or modify data and change configurations to disrupt the service. It’s therefore essential to have formally enforced mechanisms to control the allocation and use of these privileges.
2.1.3 Deliberate deletion as a result of cloud infrastructure compromise
Step to Threat Tree 4: 1.1.3
2.2 (AND) No readily recoverable backup exists on customer or 3rd party site
Replication / Backup of Data Locally (Customer): The customer can put in place mechanisms to maintain a backup or archive of data held in the cloud. Replication / Backup of Data to 3rd Party (Customer): Data can be archived, journaled or backed up to a third party service in addition to the normally expected backups on the cloud service
2.3 No readily recoverable backup exists on the cloud
2.3.1 Unrecoverable disaster event in cloud provider infrastructure
2.3.1.1 Disaster in paired site of cloud infrastructure simultaneously
Geographic separation: Ensuring that data centre locations are geographically distant so that any localised disaster event does not affect both data centres in an availability pair.
2.3.1.2 (AND) Disaster in a single cloud infrastructure location
Fire Suppression & Containment: Physical equipment to detect the presence of fire, suppress that fire and contain it to one area of a data centre for a period of time. Power Management: Uninterruptible backup power supplies capable of maintaining operations during power disturbances. Heating & Cooling: Environmental controls over heating and cooling to ensure the data centre can be maintained at a reliable working temperature. Site Selection: Due diligence around the selecting the location of a data centre to limit exposure to environmental conditions like flood, fire, or other extreme weather events as well as geopolitical and legal stability.
2.3.1.3 (AND) Disaster recovery processes fail
System Redundancy and Failover: High availability measures including redundancy and the ability to failover systems
108
Identifier Event / Condition Aligned Controls
Tested Disaster Recovery Processes: A process of regular testing of disaster recovery processes to ensure that in the event of a real disaster, the processes work.
2.3.2 Cloud provider backup or recovery processes fail
Offline Backup: This is the process of taking a backup copy of customer data at scheduled intervals, then transferring that copy to a secure location. The volume of data involved and opportunity for errors in this process has led to the decline in practices for offline backup in favour of near-continuous data replication. Data Replication: Replication of data within a data centre and between paired data centres is generally more favoured than offline backup. Tested Backup and Recovery Processes: Ensuring that backup and recovery processes are tested and that both responsibilities and procedures are understood between the customer and the cloud service provider.
109
110
Threat Tree 7: Disruption of Cloud
Service
As organisations increasingly rely on their
applications and information technology
services, it is understandable to be concerned
that disruption of a cloud service could have
significant impact on the business operation.
This third risk event focuses then on the
conditions that could result in a cloud service
becoming unavailable.
Generally, cloud service providers will have an
established Service Level Agreement defining
the expected level of service availability. This
may be backed by financial or other
compensation mechanisms if the actual
availability levels are less than the SLA.
However, it is unlikely that this compensation
will adequately reflect the actual cost of
business disruption to a customer.
There are four main circumstances considered
in this threat tree. The first scenario is that of
the service becoming unavailable due to
exceptional network or processing load on the
service. This could result from a denial of
service attack, an inadequate amount of cloud
or network resources being provisioned or
congestion and disruption on the network.
Effectively dealing with these circumstances
requires an ability for the cloud service
provider to dynamically provision and deal
with exceptional load events. It also requires
customers to have adequate network
provisioning and planning.
A second scenario is for the service to be
unavailable due to planned downtime. Large
scale cloud providers generally plan and
execute updates and changes in a manner
that avoids significant amounts of planned
downtime. Planned downtime should be
reflected in the Service Level Agreement and
care should be taken to consider the timing of
any planned events. For example, a common
planned downtime window for a US based
service is late on a Sunday night, which may
equate to late Monday afternoon in
somewhere like Australia.
The third scenario for disruption to service is a
failure event either within the cloud service,
on customer infrastructure for connecting to
the cloud or a configuration error by either
the cloud provider or the customer. Dealing
with this scenario requires a combination of
high availability technical measures within the
cloud service as well as change control
processes on both sides.
A final scenario is for the service to become
unavailable if it is terminated by the cloud
service provider, either because they’ve
become financially non-viable or have simply
chosen to terminate the service as a business
decision. Apart from contractual
commitments, this requires due diligence in
assessing the track record and viability of the
cloud service provider.
111
Identifier Event / Condition Aligned Controls
3 Disruption of cloud service
3.1 Service unavailable or degraded due to network or processing load
3.1.1 Denial of service attack Content Distribution Network: A content distribution network can help mitigate the effects of a denial of service attack by having content distributed and cached at multiple locations. This spreads the impact of an attack and enables multiple alternate channels to a cloud resource. Denial-of-Service Throttling: Effective detection, and classification of traffic into ‘good’ and ‘bad’ combined with automated mechanisms to throttle bad traffic out.
3.1.2 Inadequate cloud infrastructure resources provisioned
Dynamic cloud resource provisioning: Mechanisms to scale up and scale out on demand either automatically based on identified resource needs or on customer demand. This dynamic resource provisioning is a foundational characteristic of public cloud. Service / Resource Planning (Customer): Customers of cloud services also need to plan for the resources they will need, then establish authorisation for the cloud service provider to scale to satisfy demand.
3.1.3 Inadequate network resources provisioned
Network provisioning and scale-up (Customer): Commonly, the performance bottleneck in cloud services is the network connection between the cloud service provider and the consumer. A network connection that can be adjusted in terms of bandwidth and quality on demand is usually required.
3.1.4 Network service congested or disrupted
Multiple network routes within cloud service: Use of multiple network paths within a cloud data centre and between data centres enables a greater level of resilience to network issues. Likewise connections between data centers man need to traverse multiple paths Multiple network routes for customer (Customer): Use of multiple network routes between the customer and the cloud service provider might include for example both a direct, dedicated connection and an encrypted connection over the public internet.
3.2 Service unavailable due to planned service downtime
Contracted service level agreement: A formal agreement around service levels including availability targets, scheduled downtime and compensation for failing to achieve SLA’s. High availability configuration (Customer): The cloud customer can often make choices that increase the availability of the overall solution. For example, having multiple virtual machines live operating in different availability zones. Generally cloud service providers will schedule changes so that they only occur in one availability zone at a time.
3.3 Service unavailable due to a failure event
3.3.1 Cloud service provider hardware or software failure
3.3.1.1 (AND) Hardware or software fails Clustering and Load Balancing: Distribution of processing across a number of systems can enable seamless recovery from failure.
112
Identifier Event / Condition Aligned Controls
Automated Fail Over: Clustering is often used for processing intensive systems, while failover is more commonly implemented for database and storage resources Replication of Data: Replicating data between systems to ensure a consistent replica exists on a number of systems simultaneously is essential to effective automated failover. Replication may exist within close proximity of a system or even between geographically distant data centres. High availability configuration (Customer): The cloud customer can often make choices that increase the availability of the overall solution. For example, having multiple virtual machines live operating in different availability zones. Generally cloud service providers will schedule changes so that they only occur in one availability zone at a time.
3.3.1.2 (AND) Clustering and failover measures fail
Testing of high availability processes: Frequent, periodic testing and refinement of high availability processes is essential.
3.3.2 Customer intermediary hardware or software failure
Redundancy & fail over: Sometimes cloud services rely on particular infrastructure operated by the customer, such as for authentication or networking. If these systems are not resilient, then a failure could prevent users from accessing the services. Testing of high availability processes: Frequent, periodic testing of high availability mechanisms on the customer side is necessary.
3.3.3 Configuration error Cloud service change control processes: Formalised processes for the planning, authorisation, execution and verification of changes in a controlled manner Customer change control processes (Customer): Formalised processes for the planning, authorisation, execution and verification of changes in a controlled manner
3.4 Service unavailable due to termination of service by the cloud provider
3.4.1 Service provider becomes bankrupt or otherwise ceases to operate
Financial due diligence: Examining the financial stability and liquidity of the cloud service provider
3.4.2 Service provider makes a business decision to terminate the service
Business Practices Due Diligence: Examining the track record of the cloud service provider in launching and terminating services.
Wo
rked
Exa
mp
le
113
Worked Example The following summary risk assessment describes the Department’s assessment of risk exposure for
each of the three alternative options. Risk events where the assessed exposure is higher than the
organisation’s assessed risk tolerance are particularly highlighted.
Strategic Risk Events 1: On Premise 2: Local Hosted Cloud
3: Public Cloud
Risk Event Tolerance Impact Likelihood Exposure Likelihood Exposure Likelihood Exposure
S-Co-01: Cloud service data breach M 4 2 S 1 M 1 M
S-Tr-01: Cloud service bankruptcy S 3 1 L 2 M 1 L
S-Tr-02: Inappropriate use of customer data M 4 1 M 1 M 1 M
S-Pd-01: Termination by service provider M 2 1 VL 2 L 2 L
S-Av-01: Extended period of service disruption M 3 3 S 2 M 1 L
S-Fl-01: Service inflexibility H 1 4 M 2 VL 3 L
S-Po-01: Non-portability of data on termination S 2 2 L 2 L 2 L
S-Cf-01: Unable to initially configure to fit M 3 4 H 2 M 2 M
Operational Risk Events 1: On Premise 2: Local Hosted Cloud
3: Public Cloud
Risk Event Tolerance Impact Likelihood Exposure Likelihood Exposure Likelihood Exposure
O-Ex-01: Damage due to co-tenant activities S 1 1 VL 2 VL 2 VL
O-Ex-02: Unauthorised physical access M 3 3 S 2 M 1 L
O-Co-01: Malicious insider steals data M 3 2 M 2 M 2 M
O-Co-02: Lawful confiscation of equipment S 1 1 VL 3 L 1 VL
O-Co-03: Accidental loss or theft of backups M 3 2 M 2 M 2 M
O-Tr-01: Confused security responsibilities M 3 2 M 3 S 4 H
O-In-01: Damage to logs S 2 2 L 1 VL 1 VL
O-Lt-01: Required capacity unavailable S 2 4 S 2 L 1 VL
O-Lt-02: Network congestion H 2 3 M 2 L 2 L
O-Lt-03: Variable performance under load H 2 4 S 3 M 2 L
O-Av-01: Network disruption S 3 3 S 3 S 3 S
O-Av-02: Downtime during business hours M 3 3 S 2 M 2 M
O-Re-01: Disruption due to natural disaster S 3 2 M 1 L 1 L
O-Cf-01: Lack of skilled resources M 2 4 S 2 VL 2 VL
Wo
rked
Exa
mp
le
114
Compliance Risk Events 1: On Premise 2: Local Hosted Cloud
3: Public Cloud
Risk Event Tolerance Impact Likelihood Exposure Likelihood Exposure Likelihood Exposure
C-Co-01: Lawful warrant compelling disclosure M 2 1 VL 1 VL 2 L
C-Co-02: Leakage of sensitive data into cloud S 2 3 M 3 M 3 M
C-Tr-01: 3rd party supplier compliance issues M 3 1 L 2 M 2 M
C-Tr-02: Failure to maintain adequate controls M 4 3 H 2 S 1 M
C-Pr-01: Insider accesses private data M 3 2 M 1 L 1 L
C-Pr-02: Inadequate privacy legal protections M 2 1 VL 1 VL 3 M
C-Pr-03: Cloud practices inadequate for privacy M 3 2 M 2 M 2 M
C-Pd-01: Change in cloud provider practices S 4 1 M 2 S 2 S
C-In-01: Unable to respond to legal discovery H 2 1 VL 2 L 3 M
C-Pp-01: Insufficient disaster preparedness H 2 3 M 2 L 1 VL
C-Fl-01: Unable to meet new compliance needs S 2 2 L 2 L 3 M
C-Po-01: Contractual restriction on portability M 2 1 VL 3 M 2 L
Technical Risk Events 1: On Premise 2: Local Hosted Cloud
3: Public Cloud
Risk Event Tolerance Impact Likelihood Exposure Likelihood Exposure Likelihood Exposure
T-Ex-01: Isolation failure between tenants S 2 1 VL 2 L 1 VL
T-Ex-02: Exploitation of customer gateway M 4 2 S 2 S 2 S
T-Ex-03: Isolation failure on cloud fabric S 2 1 VL 2 L 2 L
T-Ex-04: Exploitation of cloud operations M 3 3 S 2 M 2 M
T-Ex-05: Customer infrastructure exploited M 4 3 H 3 H 3 H
T-Co-01: Interception in transit to customer M 3 2 M 1 L 1 L
T-Co-02: Data visible on reallocated resources M 3 1 L 1 L 1 L
T-Co-03: Interception between data centres M 3 1 L 1 L 2 M
T-Co-04: Interception due to loss of encryption M 4 2 S 1 M 1 M
T-In-01: Backup failure S 2 3 M 1 VL 1 VL
T-Su-01: Cascading failure in cloud H 2 1 VL 1 VL 1 VL
T-Ip-01: Failed integration of security systems H 1 1 VL 2 VL 3 L
115
Step 5 – Decide & Plan
Overall Assurance Evaluation
The final step in the assurance evaluation is to
combine all inputs and the outcomes of
functional, compliance and risk assessments
to develop an executive recommendation for
the business. This recommendation should
reflect the strategic intent, context and
assurance objectives mapped out at the
beginning of the process. And it needs to
provide sufficient guidance for an executive
decision on which of the multiple alternative
options are preferable from a security
assurance perspective.
There are three principal components to the
final recommendation. The first component
comprises an assessment of how the cloud
service can integrate with the organisation,
addressing the internal requirements around
identity & access, information protection and
management & monitoring. It aims to answer
the question: Which of the alternatives best
aligns with our organisation’s security needs?
The second part is a compliance assessment
covering privacy, records, information security
and other externally imposed compliance
requirements. This part addresses the
question: Which of the options best enables
our organisation to satisfy compliance
obligations?
The third part presents a summary of the risk
assessment across the assurance domains of
trustworthiness, resilience and adaptability
considering impacts of a strategic,
operational, compliance and technical nature.
Along with the summary, the top 5 risk events
are also described, possibly along with
guidance for their treatment.
Finally, the executive recommendation
provides a subjective overall assessment of
each solution relative to the assurance
objectives.
A template for presenting this
recommendation is shown in the worked
example.
Figure 31: Assembling the outputs from each stage of the assurance process to provide an
overview and executive recommendation
116
Ongoing Governance
The focus of this handbook has been
undertaking an assurance evaluation and then
making a decision, but assurance is not just a
point-in-time process. During the design,
deployment and operational stages of
implementing a cloud solution, there are a
number of aspects of assurance and security
governance that need to be monitored.
Engineering Quality It’s important to ensure that the engineering
plan to migrate to the cloud solution
incorporates security throughout the design
and deployment. Decisions that may have
been made during the assurance evaluation to
incorporate specific mitigations or additional
controls will need to be factored into the
design. New information and technical
challenges may emerge during the
engineering process requiring security
assessment.
User Privileges and Credential
Management The cloud service provider will generally have
formalised controls around privileges and
credentials for operator personnel, but the
cloud customer also has to establish similar
controls. Who will have administrative access
to the cloud service, capable of configuring
the service or adding new users? How will
users be authenticated and how will their
credentials be distributed? How will their
access be revoked if they leave the
organisation? These are just some of the
questions that the cloud service customer
needs to address.
Information Classification The risk and compliance assessment is based
on an understanding of the sensitivity or
classification of the customer data. For
example, the assessment may have concluded
that for confidential data, the cloud service
provides adequate protection as default, but
the controls are not yet adequate for Secret
data.
Classification is typically the responsibility of
the customer though, so it’s necessary to have
processes to actually confirm that data being
processed in this example is restricted to only
confidential data or lower, and that it doesn’t
arbitrarily expand to be used for more
sensitive data classifications without
appropriate assessment.
Operational Assurance Cloud services are always changing as new
functionality and capabilities are released.
New threats and techniques emerge to be
used by attackers. Compliance requirements
change as government and regulators adapt
to the opportunities and risks of cloud.
The customer needs to be able to validate
that these features, threats or compliance
requirements don’t materially affect their
compliance and risk assurance posture. It
may be appropriate to implement additional
compensating controls or take advantage of
new features to offset a change in assessed
chance in risk or compliance.
Monitoring & Management Processes The organisation needs to assess if it has the
processes in place to monitor and manage
their consumption of cloud services. The
technical skills involved are very different to
managing on-premise infrastructure and the
mechanisms by which services are monitored
and managed require different skills.
From a security incident response
perspective, the process between a cloud
117
service provider and a customer will be very
different. If an attack is prevented by the
cloud service provider, then the customer
may not even be notified. If an attack is
successful, then the investigation performed
by the cloud provider may result in very little
information being available to the customer.
Forensic information may be unavailable or of
a very different nature to on-premise forensic
investigations.
Wo
rked
Exa
mp
le
118
Worked Example
Executive Background
The Department has undertaken an evaluation of the security of three alternative delivery options
for the Department’s collaboration & communications modernisation program. The three options
were: to implement the solution on-premise; to source the solution from a local cloud hosting
provider; to source the solution from a global cloud service provider. It is understood that each
option comes at significantly varying costs, but cost was out-of-scope for this evaluation.
The evaluation sought to assess the capability of each option to:
Integrate with our existing systems and Departments future information security needs;
Enable compliance with the Department’s regulatory and other external requirements;
Minimise the information security risk to the Department
Fit to Internal Requirements
An assessment covering the ability of each alternative option to satisfy the Department’s
requirements in the areas of identity & access, management & monitoring and information
protection is summarised below. All solution options provide an adequate capability to integrate
with the Department’s existing systems. Public cloud and local hosted cloud provide a lesser
amount of direct control and therefore less visibility into internal security events, but both cloud
options provide much better identity & access capabilities as well as better information protection
capabilities than the Department could realistically achieve with an on-premise solution.
Identity & Access 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
Authentication
Access Control
Provisioning & Lifecycle Management
Audit & Logging
Management & Monitoring 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
Configuration Management
Security Management & Monitoring
Service Management & Monitoring
Information Protection 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
Encryption
Threat Management
Data Loss Prevention
Wo
rked
Exa
mp
le
119
Compliance Requirements
An assessment of compliance requirements covering in particular privacy, records, information
security and freedom-of-information was undertaken. This included a review of current compliance
practices, workshops with cloud service providers and a review of their draft contract terms.
Although a number of considerations were noted for consideration in contract negotiation and
detailed design, there were no areas of non-compliance identified. To emphasise in particular, this
assessment found the Department could implement the solution using an offshore public cloud
provider while in full compliance with our privacy and record-keeping regulatory requirements. The
practice of third party attestation of security in cloud service providers would help the Department
in meeting it’s information security compliance goals. The only area that may require attention is
around procurement practices as the conventional approach of tendering and fixed-price, fixed-
specification procurement will be problematic.
Compliance Area 1: On Premise 2: Local Hosted Cloud 3: Public Cloud
Records
Privacy
Information Security
Freedom-of-Information
Procurement
Risk Assessment
A risk assessment across strategic, operational, compliance and technical domains was performed
for each of the three alternative options. Risk events for which the exposure was determined to be
equivalent to or higher than the organisational tolerance are listed in the following table.
The key findings regarding risk can be summarised as:
The highest risk option is to undertake to develop the solution on premise. The
Department’s capacity to implement and monitor effective information security controls is
less than either the local hosted cloud or public cloud provider.
There is little appreciable risk differential between having the solution implemented on a
local hosted cloud versus a public cloud. The actual location of data only affects a small
number of risk events, such as interception between data centres and inadequacy of legal
privacy protections. Each of these risk events were assessed to be within the tolerance for
risk exposure.
A number of important risks were identified that highlight the need for rapid improvement
in the Department’s current practices regardless of the solution option chosen:
o Exploitation of customer gateway
o Exploitation of customer infrastructure
o Confused security responsibilities
Wo
rked
Exa
mp
le
120
Strategic Risk Events 1: On Premise 2: Local Hosted Cloud
3: Public Cloud
Risk Event Tolerance Impact Likelihood Exposure Likelihood Exposure Likelihood Exposure
S-Co-01: Cloud service data breach M 4 2 S 1 M 1 M
S-Av-01: Extended period of service disruption M 3 3 S 2 M 1 L
S-Cf-01: Unable to initially configure to fit M 3 4 H 2 M 2 M
Operational Risk Events 1: On Premise 2: Local Hosted Cloud
3: Public Cloud
Risk Event Tolerance Impact Likelihood Exposure Likelihood Exposure Likelihood Exposure
O-Ex-02: Unauthorised physical access M 3 3 S 2 M 1 L
O-Co-01: Malicious insider steals data M 3 2 M 2 M 2 M
O-Co-03: Accidental loss or theft of backups M 3 2 M 2 M 2 M
O-Tr-01: Confused security responsibilities M 3 2 M 3 S 4 H
O-Lt-01: Required capacity unavailable S 2 4 S 2 L 1 VL
O-Av-01: Network disruption S 3 3 S 3 S 3 S
O-Av-02: Downtime during business hours M 3 3 S 2 M 2 M
O-Cf-01: Lack of skilled resources M 2 4 S 2 VL 2 VL
Compliance Risk Events 1: On Premise 2: Local Hosted Cloud
3: Public Cloud
Risk Event Tolerance Impact Likelihood Exposure Likelihood Exposure Likelihood Exposure
C-Tr-01: 3rd party supplier compliance issues M 3 1 L 2 M 2 M
C-Tr-02: Failure to maintain adequate controls M 4 3 H 2 S 1 M
C-Pr-01: Insider accesses private data M 3 2 M 1 L 1 L
C-Pr-02: Inadequate privacy legal protections M 2 1 VL 1 VL 3 M
C-Pr-03: Cloud practices inadequate for privacy M 3 2 M 2 M 2 M
C-Pd-01: Change in cloud provider practices S 4 1 M 2 S 2 S
C-Po-01: Contractual restriction on portability M 2 1 VL 3 M 2 L
Technical Risk Events 1: On Premise 2: Local Hosted Cloud
3: Public Cloud
Risk Event Tolerance Impact Likelihood Exposure Likelihood Exposure Likelihood Exposure
T-Ex-02: Exploitation of customer gateway M 4 2 S 2 S 2 S
T-Ex-04: Exploitation of cloud operations M 3 3 S 2 M 2 M
T-Ex-05: Customer infrastructure exploited M 4 3 H 3 H 3 H
T-Co-01: Interception in transit to customer M 3 2 M 1 L 1 L
T-Co-03: Interception between data centres M 3 1 L 1 L 2 M
T-Co-04: Interception due to loss of encryption M 4 2 S 1 M 1 M
T-In-01: Backup failure S 2 3 M 1 VL 1 VL
Wo
rked
Exa
mp
le
121
Overall Assurance Assessment
An overall subjective assurance assessment was developed returning to the initial assurance
objectives laid out in the first stage of the evaluation. This assessment is subjective, combining all of
the inputs, discovery and findings throughout this evaluation process to provide a summary
assessment.
Objective Importance
Trustworthiness Minimal Extreme
Option 1 (on-premise) has been assessed as presenting a high risk of being unable to meet the confidentiality and exploitability assurance requirements. This is due to concerns about the ability of the Department to sufficiently secure internal systems against modern threats. However, the on-premise option offers greater transparency, privacy and predictability. Both of the cloud options are assessed to meet the assurance objectives.
Confidentiality
Integrity
Privacy
Transparency
Predictability
Exploitability
Resilience Minimal Extreme
The public cloud option achieves the highest rating in terms of resilience on all assurance criteria, closely followed by the local hosted cloud option. The on-premise deployment option achieves the lowest assurance rating in resilience. This is because the Department has significant gaps in business continuity practices and limitations in the amount of resilient infrastructure that can be implemented.
Preparedness
Responsiveness
Survivability
Durability
Availability
Load Tolerance
Adaptability Minimal Extreme
An on-premise implementation will provide the greatest level of adaptability because the Department can adapt the system in any way required. This adaptation will of course be at the cost of the Department. The public cloud solution can provide the next highest level of adaptability as the hosted cloud solution has less ability to be extended or substituted.
Portability
Configurability
Flexibility
Interoperability
Substitutability
Extensibility
Wo
rked
Exa
mp
le
122
Recommendation
The conclusion of this assurance evaluation is that Option 2 (Local Hosted Cloud) or Option 3 (Public
Cloud) can both adequately satisfy internal requirements, compliance requirements and risk
assessment. Either option is viable for the Department to proceed with, so selection may require a
consideration of cost and delivery.
The option of implementing the solution on-premise (Option 1) although fully satisfying compliance
requirements represented the least optimal fit to long term requirements and the highest overall
risk. The recommendation is not to proceed with Option 1.