cost effective it strategies to lead in a digital world - ibm€¦ · cost effective it strategies...
TRANSCRIPT
Cost Effective IT Strategies to Lead in a Digital World
Steve MillsSenior Vice President and Group ExecutiveIBM Software and Systems
October 7, 2013
© 2013 IBM Corporation
Smarter Planet Solutions Increasing Demands on IT
AT&T transfers about 30 Petabytes of data through its network daily
150 Exabytes global size of “Big Data” in Healthcare, growing between 1.2 and 2.4 EX / year
Facebook processes 500+ Terabytes of data daily
By 2016, annual Internet traffic will reach 1.3 Zettabytes
$155B Worldwide sensor market in 2011, forecasted to grow to $240B in 2016
450B Business transactions / day over the Internet by 2020
118B E-mails sent daily from a total of 3.4B e-mail accounts; volume growing to 168B in 2015
6B Mobile phones worldwide
© 2013 IBM Corporation
Five Mega-trends are Accelerating the New Era of IT
The rapid adoption of cloud computing
Business advantages created through data and analytics
Increased focus on security
Consolidation and integration of IT infrastructure
Emergence of social and mobile
© 2013 IBM Corporation
Shifting Trends in the IT Environment
From monolithic applications to dynamic servicesFrom static infrastructure to cloud services
From programmed systems to learning systems
From structured data at rest to unstructured data in motionFrom stable well-defined workloads to unpredictable workloads
From standard devices to a variety of devicesFrom proprietary standards to open innovation
© 2013 IBM Corporation
The Evolution of Today’s Data Center is Based on Four Critical Actions and Enabling Capabilities
Better Business Economics
Accelerated Business Velocity
Physical Systems
…Simplified, Automated, ProactiveComplex, Skill Intensive, Reactive Monitoring…
Federated Processes
© 2013 IBM Corporation
Workload Requirements Should Determine Which of the Available Deployment Options is Used to Maximize Business Success
Functional and Non-Functional Workload Attributes…
Deployment Speed and Agility Elasticity and Scaling Security and Compliance Reliability, Availability & Service Level Transactional Integrity Operational Cost and Efficiency Data Location Regulatory Items Resource availability Co-Location requirements Storage Access Operations Costs
Public
SaaS
Public or Private
PaaS
Public or Private
IaaS
Private
On Premise
© 2013 IBM Corporation
IT Operating CostsThe Cost of Labor Outpaces New Investment
Source: IDC, 2013
Power and cooling costs
63%8%
29%
1996
$100 B
43%11%
46%
2001
$130 B
33%16%
51%
2006
$175 B20%12%
68%
2013
$249 B est.
New Server spending
Server management and admin costs
Worldwide Spending on Servers, Power, Cooling & Management
Administration
© 2013 IBM Corporation
Increased Demands on IT Create Data Center Sprawl
Data centers have doubled their energy use in the past 5 years 18% increase in
data center energy costs projected
32.6 million servers worldwide 85% idle computer capacity 15% of servers run 24/7 without
being actively used on a daily basis
Internet connected devices growing 42% per year to 1 Trillion devices by 2017
Between 2000 and 2010 servers grew 6x storage grew 69x
Virtual machines growing 42% per year
Growing Cumulative Value of
Software Investments
© 2013 IBM Corporation
Sprawl Drives Unsustainable Costs
Cost to Manage Growth of Inventory Consumes
IT Budget
© 2013 IBM Corporation
IT Must Break Through Budget and Resource Barriers
Getting Up and Running 2-3 months to specify and procure 2-3 months to integrate, configure
and deploy
Development Operations 3-6 months to go from
development to production
Ongoing Effort1-3 months to troubleshoot and tuneOngoing effort and downtime to maintain, scale and upgrade
IT RealityBusiness Goals
Driving business innovation Make new markets Respond to competitive threats Enhance the customer
experience
Grow top and bottom line by:
Typical Results 23% of new IT projects (worldwide) deploy late 55% experience application downtime for major
infrastructure upgrades once deployed
Source: A commissioned study conducted by Forrester Consulting on behalf of IBM
© 2013 IBM Corporation
Only 1 in 5 Can Allocate More than Half Their IT Budget to Innovation
More Effective Use of Technology86% first and fast technology adoption58% move virtual machines to meet desired outcomes93% use storage virtualization87% use a storage service catalog (tiered storage)
Less Effective Use of Technology43% first and fast technology adoption1% move virtual machines to meet desired outcomes
21% use storage virtualization3% use a storage service catalog (tiered storage)
Most Efficient IT OrganizationsLess Efficient IT Organizations
Maintaining existing infrastructure
65%
New projects
35%Maintaining existing infrastructure
47%
New projects
53%
© 2013 IBM Corporation
How Do You Quantify IT Economics?
Workload Identified for
Analysis
Deployment Choices
Key Steps in Analysis
1. Establish Equivalent Configurations Needed to deliver workload
2. Compare Total Cost of Ownership TCO looks at different dimensions of cost
Deploy on other platformsDo Nothing Optimize current
environment
Application XYZ(Production, Development, QA)
Frontends
Databases
Other Components
© 2013 IBM Corporation
Understanding “Total Cost of Ownership (TCO)”Four Dimensions of Cost Should Be Considered
Cost components
Environments
Time Factors
Non-Functional Requirements / Qualities of Service
© 2013 IBM Corporation
Many Cost Components80:20 rule helps to achieve reasonable results in a short time
Hardware
Software
People
Network
Storage
Facilities
Components
Space, electricity, air cooling, infrastructure including UPS and generators, alternate site(s), bandwidth
ECKD, FBA, SAN, compressed, primary, secondary Disk (multiple vendors), tape, Virtual, SSD
Adapters, switches, routers, hubs Charges, allocated or apportioned, understood or clueless
FTE rate, in house versus contract
List versus Discounted Fully configured vs. basic, Production vs. Disaster Recovery Refresh / upgrade, Solution Edition…
IBM and ISV, OTC and annual maintenance (S&S) MLC, PVU, RVU, ELA, core, system
© 2013 IBM Corporation
Environments Multiply Components
Production/Online Batch/Failover
Quality Assurance
Disaster RecoveryTestDevelopment
Hardware
Software
People
Network
Storage
Facilities
ComponentsEnvironments
© 2013 IBM Corporation
Time Factors Drive Growth and Cost
Migration time and effort
Business organic growth and/or planned business changes affect capacity requirements
- e.g. Change of access channel or adding a new Internet accessible feature can double or triple a components workload
- Link a business metric (e.g. active customer accounts) to workload (e.g. daily transactions) and then use business inputs to drive the TCO case
Other periodic changes - hardware refresh or software remediation
© 2013 IBM Corporation
Availability … Reliability …Security … Scalability …
Qualities of Service, Non-Functional Requirements
Production/Online Batch/Failover
Quality Assurance
Disaster RecoveryTestDevelopment
Hardware
Software
People
Network
Storage
Facilities
ComponentsEnvironments
Time
Non-Functional Requirements Can Drive Additional Resource Requirements
© 2013 IBM Corporation
Total Cost of Ownership - Understanding the Complete Story
Production/Online Batch/Failover
Quality Assurance
Disaster RecoveryTestDevelopment
Hardware
Software
People
Network
Storage
Facilities
ComponentsEnvironments
Time“Qualities of Service” such as Availability,
Reliability, Security and Scalability
© 2013 IBM Corporation
Scale-up AND Scale-out
Both valid Unit of work dynamics must be understood Data access patterns must be understood Sharing and isolation patterns must be understood
Scale out Scale out
Scale up
Scale down
© 2013 IBM Corporation
Larger Servers with More Resources Make More Effective Consolidation Platforms
Most workloads experience variance in demand
When you consolidate workloads with variance on a virtualized server, the variance of the sum is less (statistical multiplexing)
The more workloads you can consolidate, the smaller is the variance of the sum
Consequently, bigger servers with capacity to run more workloads can be driven to higher average utilization levels without violating service level agreements, thereby reducing the cost per workload
© 2013 IBM Corporation
Observations
There is a benefit to large scale servers- The headroom required to accommodate variability goes up only
by sqrt(n) when n workloads are pooled
- The larger the shared processor pool is, the more statistical benefit you get
- Large scale virtualization platforms are able to consolidate large numbers of virtual machines because of this
Servers with capacity to run more workloads can be driven to higher average utilization levels without violating service level agreements
© 2013 IBM Corporation
A Single Workload Requires a Machine Capacity of 6x the Average Demand
Server Capacity Required
60/sec
Average Demand
m=10/sec
Assumes coefficient of variation = 2.5, required to meet 97.7% SLA
Server utilization = 17%
6x Peak To Average
© 2013 IBM Corporation
Consolidation of 4 Workloads Requires Server Capacity of 3.5x Average Demand
Server Capacity Required 140/sec
Average Demand
4*m=40/sec
3.5x Peak To Average
Assumes coefficient of variation = 2.5, required to meet 97.7% SLA
Server utilization = 28%
© 2013 IBM Corporation
Consolidation of 16 Workloads Requires Server Capacity of 2.25x Average Demand
Server Capacity Required 360/secAverage Demand
16*m=160/sec
Assumes coefficient of variation = 2.5, required to meet 97.7% SLA
2.25x Peak To Average
Server utilization = 44%
© 2013 IBM Corporation
Consolidation of 144 Workloads Requires Server Capacity of 1.42x Average Demand
Server Capacity Required 2,045/secAverage Demand 144*m=
1440/sec
Assumes coefficient of variation = 2.5, required to meet 97.7% SLA
1.42x Peak To Average
Server utilization = 70%
© 2013 IBM Corporation
Benefit of Statistical MultiplexingA single virtualized server with a large pool of shared processors can run more
workloads than several smaller servers with the same total number of processors
3xmore workloads 4.4x
more workloads
32 workloads1 per server
96 workloads24 per server
140 workloads140 per server
32 servers with 8 cores256 cores total1024 threads
4 servers 64 cores256 cores total1024 threads
1 server with 256 cores256 cores total1024 threads
Each workload is the largest that can fit on a
single 8 core server
© 2013 IBM Corporation
zEnterprise EC12 Efficiency at ScaleWorld Record SAP Transaction Banking
2 x 22 DB cores2 x 7CF cores
DB2 for z/OSCompetitor DB on Intel
128 DB cores
Database
8x 3850 x5 with 16 cores (dual active clusters) zEC12 2-way data sharing Sysplex
DatabaseSAP
ApplicationsSAP
Applications
Database Unit Cost (5yr TCA)$ 0.30 / Postings per Hour
Database Unit Cost (5yr TCA)$ 0.15 / Postings per Hour
World record at half
the cost
Postings per Hour 59.1 M# of Accounts 150 MDB2 Solution Ed (Hdw+Sft) $ 7.49 MCapacity Backup $ 1.24 MTotal Cost $ 8.73 M
Postings per Hour 42.0 M# of Accounts 90 MHardware $ 0.63 MSoftware $ 11.98 MTotal Cost $ 12.61 M
© 2013 IBM Corporation
zEnterprise EC12 Efficiency at ScaleConsolidated Oracle Database Workloads on zLinux
Three Database Workloads 3 Oracle RAC clusters 4 nodes per cluster Each node is a zLinux guest zEC12 with 27 IFLs*
$ 5.7 M (3yr. TCA)
Which platform provides the lowest TCA over 3 years?
Oracle DB Workload
3 Oracle RAC clusters 4 server nodes per cluster 12 total 16 core servers
(192 cores)
$ 13.2 M (3yr. TCA)
* Estimated values as of August 2012
TCA includes hardware, software, maintenance, support and subscription. Workload Equivalence derived from a proof-of-concept study conducted at a large Cooperative Bank.
Half the cost
© 2013 IBM Corporation
Better Labor Productivity - U.S. Insurance CompanyLarge Systems with Centralized Management
$ 0.12 per claim
$ 0.79 per claim
Identical function Different platforms
Mainframe support staff has6.6x better
productivity
IBM System z CICS/DB2
MIPs used for commercial claims processing production / dev / test 945
Claims per year: 4,056,000
Competitor UNIX Servers + ISV
Claims per year: 327,652
Development/Test Servers 2 Large Enterprise-class Intel
Production Servers 2 Large Enterprise-class Intel
© 2013 IBM Corporation
Data Duplication Wastes Storage and Processing Capacity
Systems of record
Local solutions for OLTP
DB
Appl
DB
Appl
DB
Appl
DB
Appl
DB
Appl
DB
Appl
IMSDB2
Oracle
A Large European Bank Proliferation of local solutions
- Applications + Databases 1,000 LPARs on 750 cores with
14,000 software titles 120 database images Heavy data movement
- Bulk data transfers (ETL) to local DB
ETL consumed 28% of distributed cores and 16% of MIPS
A Large Asian Bank ETL consumed 8% of total distributed core
and 18% of total MIPS
© 2013 IBM Corporation
Optimize Server Utilization
Reduce Server Sprawl
Consolidate Servers Server Virtualization Reduce # of Data Centers Asset Management
Manage Workloads
Workload Optimized Systems …. Best Fit
Automate Processes
Automation Provisioning Workload Scheduling
Optimize Service Delivery
Integrated Service Mgmt. (ISM) Self-provisioned Cloud Services
© 2013 IBM Corporation
Storage Efficiency and Best Practices
Stop Storing So Much
Data Compression Data Deduplication
Store More With What’s On the Floor
Storage Virtualization Thin Provisioning Consolidated Storage Management
Move Data to the Right Place
Automated Tiering Automated Data Migration Policy-based Management
© 2013 IBM Corporation
Eliminate Redundant Software and Data
Applications Identify and Eliminate Redundant Systems & Applications Reuse of Common Services - SOA
Largest Cost Savings but Slower to Achieve
Data
Reduce Database Size - Data Compression Reduce Multiple Copies - Data De-duplication Information Integration - Master Data Management Data Governance - Archiving
© 2013 IBM Corporation
Improve Service Delivery
Integrated Service Management (ISM) Visibility, Control, Automation
Transform to Private Cloud Delivery Model Self-service Automated Provisioning
and Workload Scheduling Elastic Capability to Expand Pay-as-you-go
Integrate with Public Clouds
Improves Labor Productivity and Flexibility
Integrated Service Management
Visibility AutomationControl
© 2013 IBM Corporation
Workload Requirements Should Determine Which of the Available Deployment Options is Used to Maximize Business Success
Functional and Non-Functional Workload Attributes…
Deployment Speed and Agility Elasticity and Scaling Security and Compliance Reliability, Availability & Service Level Transactional Integrity Operational Cost and Efficiency Data Location Regulatory Items Resource availability Co-Location requirements Storage Access Operations Costs
Public
SaaS
Public or Private
PaaS
Public or Private
IaaS
Private
On Premise
© 2013 IBM Corporation
On-Premise or Public Cloud: The Choice is Yours
On-Premise Public Cloud
Greater control More security Higher performance Deeper compliance Customizable
May be higher cost On-site maintenance Capacity ceiling
Simplicity and efficiency May be lower cost Reduced time No maintenance Pay-as-you-go model
Lack of control Scaling issues Lack of investment Perceived weaker security
Dis
adva
nta
ges
A
dvan
tage
s
Public Cloud
© 2013 IBM Corporation
Significant variation in demand for IT resources
- Exploit the differential and reduce operating costs by only matching supply to the level of demand
- Metered billing can allow resources to scale up/down to meet demand
To realize this benefit- Peak/average demand ratio must be greater than the cost ratio
of Public Cloud over On-Premise Hosting- Application must be able to scale horizontally- Existence of sufficient density of compute nodes to allow scaling
When Is Public Cloud Hosting Cost Effective
© 2013 IBM Corporation
Calculating Comparative Costs CAPEX vs. OPEX Bulk of costs over the lifetime of an application are not related
to the purchase of physical infrastructure; therefore, Total Cost of Ownershipis more meaningful in comparing On-Premise versus Public Cloud costs
Tangible Assets- Server Hardware- Network Hardware- Storage Hardware
Intangible Assets- Hardware Maintenance Contracts- Software License Contracts- Software Support Contracts
Utilities
Real Estate
Human Labor- Data Center Operators / Administrators- Network and Storage Administrators- Systems Administrators- Software Developers / Architects- Help Desk Operatives- General & Administrative- Managerial
Downtime / Outages
Auditing / Compliance
© 2013 IBM Corporation
Additional Intangible Costs Need to be Considered
Transformation costs in moving Application
Reliability and availability
Data Center and Operations quality
Extensibility
Security and Privacy
Scalability
Capacity
© 2013 IBM Corporation
Cloud Economics
Cloud services are complex and varied and will continue to evolve
Cost savings from the use of Public Cloud services is highly dependent upon the type of application / workload, its suitability to Cloud, and the level(s) of efficiency and optimization of on-premise IT resources and operations
Evaluation of the costs associated with each application is required to gain a true understanding of the comparative costs
Transformation costs associated with migrating an application from on-premise to Cloud must be factored in the cost equation
The decision on the use of Public Cloud often comes down to cost versus control
© 2013 IBM Corporation
Cost Effective IT
Do not confuse price with cost
Budgeting and charge back techniques can cause false economics
Technology is a tool, not a religion …… insist on fact based analysis
TCO cannot be overlooked but neither can agility and effectiveness
Elimination is the clearest path to saving money