Cloud Forward2015

Augusto Ciuffoletti

Cloud Forward 2015Automated deployment of a microservice-based

monitoring infrastructure

Augusto Ciuffoletti

6 ottobre 2015

Cloud Forward2015

Augusto Ciuffoletti

Introducing two topics

I The title:Automated deployment of a microservice-basedmonitoring infrastructure

I Microservices

I Monitoring on demand

Let us see how the two stories merge...

Cloud Forward2015

Augusto Ciuffoletti

Introducing two topics

I The title:Automated deployment of a microservice-basedmonitoring infrastructure

I Microservices

I Monitoring on demand

Let us see how the two stories merge...

Cloud Forward2015

Augusto Ciuffoletti

Introducing two topics

I The title:Automated deployment of a microservice-basedmonitoring infrastructure

I Microservices

I Monitoring on demand

Let us see how the two stories merge...

Cloud Forward2015

Augusto Ciuffoletti

Introducing two topics

I The title:Automated deployment of a microservice-basedmonitoring infrastructure

I Microservices

I Monitoring on demand

Let us see how the two stories merge...

Cloud Forward2015

Augusto Ciuffoletti

Number one: Microservices

I A design paradigm for distributed systemI a book in O’Reilly "animal series" by S Newman

I Principles:I each component in the system is designed to provide

one small, well defined serviceI each component is a stand alone entity that interacts

with others across a network with a well definedinterface

Cloud Forward2015

Augusto Ciuffoletti

Number one: Microservices

I A design paradigm for distributed systemI a book in O’Reilly "animal series" by S Newman

I Principles:I each component in the system is designed to provide

one small, well defined serviceI each component is a stand alone entity that interacts

with others across a network with a well definedinterface

Cloud Forward2015

Augusto Ciuffoletti

Number one: Microservices

I A design paradigm for distributed systemI a book in O’Reilly "animal series" by S Newman

I Principles:I each component in the system is designed to provide

one small, well defined serviceI each component is a stand alone entity that interacts

with others across a network with a well definedinterface

Cloud Forward2015

Augusto Ciuffoletti

Number one: Microservices

I A design paradigm for distributed systemI a book in O’Reilly "animal series" by S Newman

I Principles:I each component in the system is designed to provide

one small, well defined serviceI each component is a stand alone entity that interacts

with others across a network with a well definedinterface

Cloud Forward2015

Augusto Ciuffoletti

Number one: Microservices

I A design paradigm for distributed systemI a book in O’Reilly "animal series" by S Newman

I Principles:I each component in the system is designed to provide

one small, well defined serviceI each component is a stand alone entity that interacts

with others across a network with a well definedinterface

Cloud Forward2015

Augusto Ciuffoletti

Reasons to adopt the microservicesparadigm

I simplifies maintenanceI e.g., upgrade one single component

I agility in deploymentI e.g., to scale up or down

I each component may use a different technologyI e.g., for technical or performance reasons

I simplifies developmentI e.g., each component developed by a distinct team

I robustnessI e.g., if one component fails there is a chance that the

system still works

Cloud Forward2015

Augusto Ciuffoletti

Reasons to adopt the microservicesparadigm

I simplifies maintenanceI e.g., upgrade one single component

I agility in deploymentI e.g., to scale up or down

I each component may use a different technologyI e.g., for technical or performance reasons

I simplifies developmentI e.g., each component developed by a distinct team

I robustnessI e.g., if one component fails there is a chance that the

system still works

Cloud Forward2015

Augusto Ciuffoletti

Reasons to adopt the microservicesparadigm

I simplifies maintenanceI e.g., upgrade one single component

I agility in deploymentI e.g., to scale up or down

I each component may use a different technologyI e.g., for technical or performance reasons

I simplifies developmentI e.g., each component developed by a distinct team

I robustnessI e.g., if one component fails there is a chance that the

system still works

Cloud Forward2015

Augusto Ciuffoletti

Reasons to adopt the microservicesparadigm

I simplifies maintenanceI e.g., upgrade one single component

I agility in deploymentI e.g., to scale up or down

I each component may use a different technologyI e.g., for technical or performance reasons

I simplifies developmentI e.g., each component developed by a distinct team

I robustnessI e.g., if one component fails there is a chance that the

system still works

Cloud Forward2015

Augusto Ciuffoletti

Reasons to adopt the microservicesparadigm

I simplifies maintenanceI e.g., upgrade one single component

I agility in deploymentI e.g., to scale up or down

I each component may use a different technologyI e.g., for technical or performance reasons

I simplifies developmentI e.g., each component developed by a distinct team

I robustnessI e.g., if one component fails there is a chance that the

system still works

Cloud Forward2015

Augusto Ciuffoletti

Reasons to adopt the microservicesparadigm

I simplifies maintenanceI e.g., upgrade one single component

I agility in deploymentI e.g., to scale up or down

I each component may use a different technologyI e.g., for technical or performance reasons

I simplifies developmentI e.g., each component developed by a distinct team

I robustnessI e.g., if one component fails there is a chance that the

system still works

Cloud Forward2015

Augusto Ciuffoletti

Reasons to adopt the microservicesparadigm

I simplifies maintenanceI e.g., upgrade one single component

I agility in deploymentI e.g., to scale up or down

I each component may use a different technologyI e.g., for technical or performance reasons

I simplifies developmentI e.g., each component developed by a distinct team

I robustnessI e.g., if one component fails there is a chance that the

system still works

Cloud Forward2015

Augusto Ciuffoletti

Reasons to adopt the microservicesparadigm

I simplifies maintenanceI e.g., upgrade one single component

I agility in deploymentI e.g., to scale up or down

I each component may use a different technologyI e.g., for technical or performance reasons

I simplifies developmentI e.g., each component developed by a distinct team

I robustnessI e.g., if one component fails there is a chance that the

system still works

Cloud Forward2015

Augusto Ciuffoletti

Reasons to adopt the microservicesparadigm

I simplifies maintenanceI e.g., upgrade one single component

I agility in deploymentI e.g., to scale up or down

I each component may use a different technologyI e.g., for technical or performance reasons

I simplifies developmentI e.g., each component developed by a distinct team

I robustnessI e.g., if one component fails there is a chance that the

system still works

Cloud Forward2015

Augusto Ciuffoletti

Reasons to adopt the microservicesparadigm

I simplifies maintenanceI e.g., upgrade one single component

I agility in deploymentI e.g., to scale up or down

I each component may use a different technologyI e.g., for technical or performance reasons

I simplifies developmentI e.g., each component developed by a distinct team

I robustnessI e.g., if one component fails there is a chance that the

system still works

Cloud Forward2015

Augusto Ciuffoletti

Number two: Cloud monitoring

A cloud user wants to have a functional feedback fromcloud sourced resources:

I not necessarily to verify service qualityI control a scalable resource,I provide feedback to the users,I trigger compensating actions

I NIST indicates monitoring as one of the distinctivefeatures of cloud computing

Cloud Forward2015

Augusto Ciuffoletti

Number two: Cloud monitoring

A cloud user wants to have a functional feedback fromcloud sourced resources:

I not necessarily to verify service qualityI control a scalable resource,I provide feedback to the users,I trigger compensating actions

I NIST indicates monitoring as one of the distinctivefeatures of cloud computing

Cloud Forward2015

Augusto Ciuffoletti

Number two: Cloud monitoring

A cloud user wants to have a functional feedback fromcloud sourced resources:

I not necessarily to verify service qualityI control a scalable resource,I provide feedback to the users,I trigger compensating actions

I NIST indicates monitoring as one of the distinctivefeatures of cloud computing

Cloud Forward2015

Augusto Ciuffoletti

Number two: Cloud monitoring

A cloud user wants to have a functional feedback fromcloud sourced resources:

I not necessarily to verify service qualityI control a scalable resource,I provide feedback to the users,I trigger compensating actions

I NIST indicates monitoring as one of the distinctivefeatures of cloud computing

Cloud Forward2015

Augusto Ciuffoletti

Number two: Cloud monitoring

A cloud user wants to have a functional feedback fromcloud sourced resources:

I not necessarily to verify service qualityI control a scalable resource,I provide feedback to the users,I trigger compensating actions

I NIST indicates monitoring as one of the distinctivefeatures of cloud computing

Cloud Forward2015

Augusto Ciuffoletti

Number two: Cloud monitoring

A cloud user wants to have a functional feedback fromcloud sourced resources:

I not necessarily to verify service qualityI control a scalable resource,I provide feedback to the users,I trigger compensating actions

I NIST indicates monitoring as one of the distinctivefeatures of cloud computing

Cloud Forward2015

Augusto Ciuffoletti

Our option: on demand monitoring

I Provide monitoring as part of the serviceI Give the user wide possibilities to configure a

monitoring infrastructureI Which metrics are captured and how data are

preprocessed and retrievedI Scale from simple to complex infrastructures

I Do not overkill the problem in simple casesI Cope with complex infrastructures

I Resource agnosticI Basic functionalities and unlimited pluggable

extensions

Cloud Forward2015

Augusto Ciuffoletti

Our option: on demand monitoring

I Provide monitoring as part of the serviceI Give the user wide possibilities to configure a

monitoring infrastructureI Which metrics are captured and how data are

preprocessed and retrievedI Scale from simple to complex infrastructures

I Do not overkill the problem in simple casesI Cope with complex infrastructures

I Resource agnosticI Basic functionalities and unlimited pluggable

extensions

Cloud Forward2015

Augusto Ciuffoletti

Our option: on demand monitoring

I Provide monitoring as part of the serviceI Give the user wide possibilities to configure a

monitoring infrastructureI Which metrics are captured and how data are

preprocessed and retrievedI Scale from simple to complex infrastructures

I Do not overkill the problem in simple casesI Cope with complex infrastructures

I Resource agnosticI Basic functionalities and unlimited pluggable

extensions

Cloud Forward2015

Augusto Ciuffoletti

Our option: on demand monitoring

I Provide monitoring as part of the serviceI Give the user wide possibilities to configure a

monitoring infrastructureI Which metrics are captured and how data are

preprocessed and retrievedI Scale from simple to complex infrastructures

I Do not overkill the problem in simple casesI Cope with complex infrastructures

I Resource agnosticI Basic functionalities and unlimited pluggable

extensions

Cloud Forward2015

Augusto Ciuffoletti

Our option: on demand monitoring

I Provide monitoring as part of the serviceI Give the user wide possibilities to configure a

monitoring infrastructureI Which metrics are captured and how data are

preprocessed and retrievedI Scale from simple to complex infrastructures

I Do not overkill the problem in simple casesI Cope with complex infrastructures

I Resource agnosticI Basic functionalities and unlimited pluggable

extensions

Cloud Forward2015

Augusto Ciuffoletti

Our option: on demand monitoring

I Provide monitoring as part of the serviceI Give the user wide possibilities to configure a

monitoring infrastructureI Which metrics are captured and how data are

preprocessed and retrievedI Scale from simple to complex infrastructures

I Do not overkill the problem in simple casesI Cope with complex infrastructures

I Resource agnosticI Basic functionalities and unlimited pluggable

extensions

Cloud Forward2015

Augusto Ciuffoletti

Our option: on demand monitoring

I Provide monitoring as part of the serviceI Give the user wide possibilities to configure a

monitoring infrastructureI Which metrics are captured and how data are

preprocessed and retrievedI Scale from simple to complex infrastructures

I Do not overkill the problem in simple casesI Cope with complex infrastructures

I Resource agnosticI Basic functionalities and unlimited pluggable

extensions

Cloud Forward2015

Augusto Ciuffoletti

Our option: on demand monitoring

I Provide monitoring as part of the serviceI Give the user wide possibilities to configure a

monitoring infrastructureI Which metrics are captured and how data are

preprocessed and retrievedI Scale from simple to complex infrastructures

I Do not overkill the problem in simple casesI Cope with complex infrastructures

I Resource agnosticI Basic functionalities and unlimited pluggable

extensions

Cloud Forward2015

Augusto Ciuffoletti

Find a match

I Monitoring is by nature split into small components(remember Nagios)

I monitoring probes are small components, possiblyembedded

I monitoring data crosses a pipe of processors(anonymization, aggregation etc)

I data is finally published using an endpoint reachablefrom the outside (database, web service)

I Each component is supported by a specifictechnology

I e.g., network monitoring vs storage monitoringI The on demand nature requires agility in deployment

I the cloud user that obtains a new resource may wantto monitor it

There is a match betweenmicroservices and on demand monitoring

Cloud Forward2015

Augusto Ciuffoletti

Find a match

I Monitoring is by nature split into small components(remember Nagios)

I monitoring probes are small components, possiblyembedded

I monitoring data crosses a pipe of processors(anonymization, aggregation etc)

I data is finally published using an endpoint reachablefrom the outside (database, web service)

I Each component is supported by a specifictechnology

I e.g., network monitoring vs storage monitoringI The on demand nature requires agility in deployment

I the cloud user that obtains a new resource may wantto monitor it

There is a match betweenmicroservices and on demand monitoring

Cloud Forward2015

Augusto Ciuffoletti

Find a match

I Monitoring is by nature split into small components(remember Nagios)

I monitoring probes are small components, possiblyembedded

I monitoring data crosses a pipe of processors(anonymization, aggregation etc)

I data is finally published using an endpoint reachablefrom the outside (database, web service)

I Each component is supported by a specifictechnology

I e.g., network monitoring vs storage monitoringI The on demand nature requires agility in deployment

I the cloud user that obtains a new resource may wantto monitor it

There is a match betweenmicroservices and on demand monitoring

Cloud Forward2015

Augusto Ciuffoletti

Find a match

I Monitoring is by nature split into small components(remember Nagios)

I monitoring probes are small components, possiblyembedded

I monitoring data crosses a pipe of processors(anonymization, aggregation etc)

I data is finally published using an endpoint reachablefrom the outside (database, web service)

I Each component is supported by a specifictechnology

I e.g., network monitoring vs storage monitoringI The on demand nature requires agility in deployment

I the cloud user that obtains a new resource may wantto monitor it

There is a match betweenmicroservices and on demand monitoring

Cloud Forward2015

Augusto Ciuffoletti

Find a match

I Monitoring is by nature split into small components(remember Nagios)

I monitoring probes are small components, possiblyembedded

I monitoring data crosses a pipe of processors(anonymization, aggregation etc)

I data is finally published using an endpoint reachablefrom the outside (database, web service)

I Each component is supported by a specifictechnology

I e.g., network monitoring vs storage monitoringI The on demand nature requires agility in deployment

I the cloud user that obtains a new resource may wantto monitor it

There is a match betweenmicroservices and on demand monitoring

Cloud Forward2015

Augusto Ciuffoletti

Find a match

I Monitoring is by nature split into small components(remember Nagios)

I monitoring probes are small components, possiblyembedded

I monitoring data crosses a pipe of processors(anonymization, aggregation etc)

I data is finally published using an endpoint reachablefrom the outside (database, web service)

I Each component is supported by a specifictechnology

I e.g., network monitoring vs storage monitoringI The on demand nature requires agility in deployment

I the cloud user that obtains a new resource may wantto monitor it

There is a match betweenmicroservices and on demand monitoring

Cloud Forward2015

Augusto Ciuffoletti

Find a match

I Monitoring is by nature split into small components(remember Nagios)

I monitoring probes are small components, possiblyembedded

I monitoring data crosses a pipe of processors(anonymization, aggregation etc)

I data is finally published using an endpoint reachablefrom the outside (database, web service)

I Each component is supported by a specifictechnology

I e.g., network monitoring vs storage monitoringI The on demand nature requires agility in deployment

I the cloud user that obtains a new resource may wantto monitor it

There is a match betweenmicroservices and on demand monitoring

Cloud Forward2015

Augusto Ciuffoletti

Find a match

I Monitoring is by nature split into small components(remember Nagios)

I monitoring probes are small components, possiblyembedded

I monitoring data crosses a pipe of processors(anonymization, aggregation etc)

I data is finally published using an endpoint reachablefrom the outside (database, web service)

I Each component is supported by a specifictechnology

I e.g., network monitoring vs storage monitoringI The on demand nature requires agility in deployment

I the cloud user that obtains a new resource may wantto monitor it

There is a match betweenmicroservices and on demand monitoring

Cloud Forward2015

Augusto Ciuffoletti

Find a match

I Monitoring is by nature split into small components(remember Nagios)

I monitoring probes are small components, possiblyembedded

I monitoring data crosses a pipe of processors(anonymization, aggregation etc)

I data is finally published using an endpoint reachablefrom the outside (database, web service)

I Each component is supported by a specifictechnology

I e.g., network monitoring vs storage monitoringI The on demand nature requires agility in deployment

I the cloud user that obtains a new resource may wantto monitor it

There is a match betweenmicroservices and on demand monitoring

Cloud Forward2015

Augusto Ciuffoletti

Summarizing: a IaaS provisioning

I IaaS is just for the example: could be PaaS oranything else

Cloud Forward2015

Augusto Ciuffoletti

A monitoring infrastructure

I Adding a monitoring infrastructure:I probes that collect monitoring data (collectors)I a device that processes monitoring data (sensor)

Cloud Forward2015

Augusto Ciuffoletti

A monitoring infrastructure

I Adding a monitoring infrastructure:I probes that collect monitoring data (collectors)I a device that processes monitoring data (sensor)

Cloud Forward2015

Augusto Ciuffoletti

A monitoring infrastructure

I Adding a monitoring infrastructure:I probes that collect monitoring data (collectors)I a device that processes monitoring data (sensor)

Cloud Forward2015

Augusto Ciuffoletti

A cloud interface

I we need to design an interfaceI an open standard exists: OCCI

Cloud Forward2015

Augusto Ciuffoletti

A cloud interface

I we need to design an interfaceI an open standard exists: OCCI

Cloud Forward2015

Augusto Ciuffoletti

Open Cloud Computing Interface basics

I The interface is REST, therefore web-orientedI The items accessed across the interface are entitiesI One type of entity is the resourceI Another is the link, that connects two resourcesI A type is characterized by standard features of the

instancesI attributes whose values define instancesI actions that model dynamic change

I The interface is extensible:I a type can be subtyped, thus adding new attributes

to the standard onesI an instance can be modified using mixins

Cloud Forward2015

Augusto Ciuffoletti

Open Cloud Computing Interface basics

I The interface is REST, therefore web-orientedI The items accessed across the interface are entitiesI One type of entity is the resourceI Another is the link, that connects two resourcesI A type is characterized by standard features of the

instancesI attributes whose values define instancesI actions that model dynamic change

I The interface is extensible:I a type can be subtyped, thus adding new attributes

to the standard onesI an instance can be modified using mixins

Cloud Forward2015

Augusto Ciuffoletti

Open Cloud Computing Interface basics

I The interface is REST, therefore web-orientedI The items accessed across the interface are entitiesI One type of entity is the resourceI Another is the link, that connects two resourcesI A type is characterized by standard features of the

instancesI attributes whose values define instancesI actions that model dynamic change

I The interface is extensible:I a type can be subtyped, thus adding new attributes

to the standard onesI an instance can be modified using mixins

Cloud Forward2015

Augusto Ciuffoletti

Open Cloud Computing Interface basics

I The interface is REST, therefore web-orientedI The items accessed across the interface are entitiesI One type of entity is the resourceI Another is the link, that connects two resourcesI A type is characterized by standard features of the

instancesI attributes whose values define instancesI actions that model dynamic change

I The interface is extensible:I a type can be subtyped, thus adding new attributes

to the standard onesI an instance can be modified using mixins

Cloud Forward2015

Augusto Ciuffoletti

Open Cloud Computing Interface basics

I The interface is REST, therefore web-orientedI The items accessed across the interface are entitiesI One type of entity is the resourceI Another is the link, that connects two resourcesI A type is characterized by standard features of the

instancesI attributes whose values define instancesI actions that model dynamic change

I The interface is extensible:I a type can be subtyped, thus adding new attributes

to the standard onesI an instance can be modified using mixins

Cloud Forward2015

Augusto Ciuffoletti

Open Cloud Computing Interface basics

I The interface is REST, therefore web-orientedI The items accessed across the interface are entitiesI One type of entity is the resourceI Another is the link, that connects two resourcesI A type is characterized by standard features of the

instancesI attributes whose values define instancesI actions that model dynamic change

I The interface is extensible:I a type can be subtyped, thus adding new attributes

to the standard onesI an instance can be modified using mixins

Cloud Forward2015

Augusto Ciuffoletti

Open Cloud Computing Interface basics

I The interface is REST, therefore web-orientedI The items accessed across the interface are entitiesI One type of entity is the resourceI Another is the link, that connects two resourcesI A type is characterized by standard features of the

instancesI attributes whose values define instancesI actions that model dynamic change

I The interface is extensible:I a type can be subtyped, thus adding new attributes

to the standard onesI an instance can be modified using mixins

Cloud Forward2015

Augusto Ciuffoletti

Open Cloud Computing Interface basics

I The interface is REST, therefore web-orientedI The items accessed across the interface are entitiesI One type of entity is the resourceI Another is the link, that connects two resourcesI A type is characterized by standard features of the

instancesI attributes whose values define instancesI actions that model dynamic change

I The interface is extensible:I a type can be subtyped, thus adding new attributes

to the standard onesI an instance can be modified using mixins

Cloud Forward2015

Augusto Ciuffoletti

Open Cloud Computing Interface basics

I The interface is REST, therefore web-orientedI The items accessed across the interface are entitiesI One type of entity is the resourceI Another is the link, that connects two resourcesI A type is characterized by standard features of the

instancesI attributes whose values define instancesI actions that model dynamic change

I The interface is extensible:I a type can be subtyped, thus adding new attributes

to the standard onesI an instance can be modified using mixins

Cloud Forward2015

Augusto Ciuffoletti

Open Cloud Computing Interface basics

I The interface is REST, therefore web-orientedI The items accessed across the interface are entitiesI One type of entity is the resourceI Another is the link, that connects two resourcesI A type is characterized by standard features of the

instancesI attributes whose values define instancesI actions that model dynamic change

I The interface is extensible:I a type can be subtyped, thus adding new attributes

to the standard onesI an instance can be modified using mixins

Cloud Forward2015

Augusto Ciuffoletti

An OCCI model for monitoring

I A Sensor is a subtype of the Resource typeI A Collector is a subtype of the Link typeI Add Mixins to specify the type of activity

I Legenda:I the sensor (red) is an OCCI resourceI the collectors (blue) are OCCI linksI computing boxes and the network are OCCI

resources too

Cloud Forward2015

Augusto Ciuffoletti

An OCCI model for monitoring

I A Sensor is a subtype of the Resource typeI A Collector is a subtype of the Link typeI Add Mixins to specify the type of activity

I Legenda:I the sensor (red) is an OCCI resourceI the collectors (blue) are OCCI linksI computing boxes and the network are OCCI

resources too

Cloud Forward2015

Augusto Ciuffoletti

An OCCI model for monitoring

I A Sensor is a subtype of the Resource typeI A Collector is a subtype of the Link typeI Add Mixins to specify the type of activity

I Legenda:I the sensor (red) is an OCCI resourceI the collectors (blue) are OCCI linksI computing boxes and the network are OCCI

resources too

Cloud Forward2015

Augusto Ciuffoletti

An OCCI model for monitoring

I A Sensor is a subtype of the Resource typeI A Collector is a subtype of the Link typeI Add Mixins to specify the type of activity

I Legenda:I the sensor (red) is an OCCI resourceI the collectors (blue) are OCCI linksI computing boxes and the network are OCCI

resources too

Cloud Forward2015

Augusto Ciuffoletti

An OCCI model for monitoring

I A Sensor is a subtype of the Resource typeI A Collector is a subtype of the Link typeI Add Mixins to specify the type of activity

I Legenda:I the sensor (red) is an OCCI resourceI the collectors (blue) are OCCI linksI computing boxes and the network are OCCI

resources too

Cloud Forward2015

Augusto Ciuffoletti

An OCCI model for monitoring

I A Sensor is a subtype of the Resource typeI A Collector is a subtype of the Link typeI Add Mixins to specify the type of activity

I Legenda:I the sensor (red) is an OCCI resourceI the collectors (blue) are OCCI linksI computing boxes and the network are OCCI

resources too

Cloud Forward2015

Augusto Ciuffoletti

An OCCI model for monitoring

I A Sensor is a subtype of the Resource typeI A Collector is a subtype of the Link typeI Add Mixins to specify the type of activity

I Legenda:I the sensor (red) is an OCCI resourceI the collectors (blue) are OCCI linksI computing boxes and the network are OCCI

resources too

Cloud Forward2015

Augusto Ciuffoletti

An OCCI model for monitoring

I A Sensor is a subtype of the Resource typeI A Collector is a subtype of the Link typeI Add Mixins to specify the type of activity

I Legenda:I the sensor (red) is an OCCI resourceI the collectors (blue) are OCCI linksI computing boxes and the network are OCCI

resources too

Cloud Forward2015

Augusto Ciuffoletti

How do we do that?

I We want to study the big arrow in the figureHow do we implement a monitoring infrastructure starting

form OCCI entities

Cloud Forward2015

Augusto Ciuffoletti

How do we do that?

I We want to study the big arrow in the figureHow do we implement a monitoring infrastructure starting

form OCCI entities

Cloud Forward2015

Augusto Ciuffoletti

Vintage programming but...

I For an early prototype we selected seasonedtechnologies: Java/Unix sockets/Unix Pipes

I not bound to specific programming toolsI better solutions do exist

I As a virtualization platform we selected DockerI demo is easely reproduceable/shareableI efficient to run on a single laptopI upgradeable to a real deployment

Cloud Forward2015

Augusto Ciuffoletti

Vintage programming but...

I For an early prototype we selected seasonedtechnologies: Java/Unix sockets/Unix Pipes

I not bound to specific programming toolsI better solutions do exist

I As a virtualization platform we selected DockerI demo is easely reproduceable/shareableI efficient to run on a single laptopI upgradeable to a real deployment

Cloud Forward2015

Augusto Ciuffoletti

Vintage programming but...

I For an early prototype we selected seasonedtechnologies: Java/Unix sockets/Unix Pipes

I not bound to specific programming toolsI better solutions do exist

I As a virtualization platform we selected DockerI demo is easely reproduceable/shareableI efficient to run on a single laptopI upgradeable to a real deployment

Cloud Forward2015

Augusto Ciuffoletti

Vintage programming but...

I For an early prototype we selected seasonedtechnologies: Java/Unix sockets/Unix Pipes

I not bound to specific programming toolsI better solutions do exist

I As a virtualization platform we selected DockerI demo is easely reproduceable/shareableI efficient to run on a single laptopI upgradeable to a real deployment

Cloud Forward2015

Augusto Ciuffoletti

Vintage programming but...

I For an early prototype we selected seasonedtechnologies: Java/Unix sockets/Unix Pipes

I not bound to specific programming toolsI better solutions do exist

I As a virtualization platform we selected DockerI demo is easely reproduceable/shareableI efficient to run on a single laptopI upgradeable to a real deployment

Cloud Forward2015

Augusto Ciuffoletti

Vintage programming but...

I For an early prototype we selected seasonedtechnologies: Java/Unix sockets/Unix Pipes

I not bound to specific programming toolsI better solutions do exist

I As a virtualization platform we selected DockerI demo is easely reproduceable/shareableI efficient to run on a single laptopI upgradeable to a real deployment

Cloud Forward2015

Augusto Ciuffoletti

Vintage programming but...

I For an early prototype we selected seasonedtechnologies: Java/Unix sockets/Unix Pipes

I not bound to specific programming toolsI better solutions do exist

I As a virtualization platform we selected DockerI demo is easely reproduceable/shareableI efficient to run on a single laptopI upgradeable to a real deployment

Cloud Forward2015

Augusto Ciuffoletti

Deploying a demo infrastructure

I The minimal architecture (4 dockers):I one HTTP server to host the OCCI entities

descriptionsI one dashboard to display monitoring dataI one monitored compute resourceI one sensor resource

I In OCCI viewI One compute resourceI One sensor resourceI One collector link

Cloud Forward2015

Augusto Ciuffoletti

Deploying a demo infrastructure

I The minimal architecture (4 dockers):I one HTTP server to host the OCCI entities

descriptionsI one dashboard to display monitoring dataI one monitored compute resourceI one sensor resource

I In OCCI viewI One compute resourceI One sensor resourceI One collector link

Cloud Forward2015

Augusto Ciuffoletti

Deploying a demo infrastructure

I The minimal architecture (4 dockers):I one HTTP server to host the OCCI entities

descriptionsI one dashboard to display monitoring dataI one monitored compute resourceI one sensor resource

I In OCCI viewI One compute resourceI One sensor resourceI One collector link

Cloud Forward2015

Augusto Ciuffoletti

Deploying a demo infrastructure

I The minimal architecture (4 dockers):I one HTTP server to host the OCCI entities

descriptionsI one dashboard to display monitoring dataI one monitored compute resourceI one sensor resource

I In OCCI viewI One compute resourceI One sensor resourceI One collector link

Cloud Forward2015

Augusto Ciuffoletti

Deploying a demo infrastructure

I The minimal architecture (4 dockers):I one HTTP server to host the OCCI entities

descriptionsI one dashboard to display monitoring dataI one monitored compute resourceI one sensor resource

I In OCCI viewI One compute resourceI One sensor resourceI One collector link

Cloud Forward2015

Augusto Ciuffoletti

Deploying a demo infrastructure

I The minimal architecture (4 dockers):I one HTTP server to host the OCCI entities

descriptionsI one dashboard to display monitoring dataI one monitored compute resourceI one sensor resource

I In OCCI viewI One compute resourceI One sensor resourceI One collector link

Cloud Forward2015

Augusto Ciuffoletti

Deploying a demo infrastructure

I The minimal architecture (4 dockers):I one HTTP server to host the OCCI entities

descriptionsI one dashboard to display monitoring dataI one monitored compute resourceI one sensor resource

I In OCCI viewI One compute resourceI One sensor resourceI One collector link

Cloud Forward2015

Augusto Ciuffoletti

Deploying a demo infrastructure

I The minimal architecture (4 dockers):I one HTTP server to host the OCCI entities

descriptionsI one dashboard to display monitoring dataI one monitored compute resourceI one sensor resource

I In OCCI viewI One compute resourceI One sensor resourceI One collector link

Cloud Forward2015

Augusto Ciuffoletti

Deploying a demo infrastructure

I The minimal architecture (4 dockers):I one HTTP server to host the OCCI entities

descriptionsI one dashboard to display monitoring dataI one monitored compute resourceI one sensor resource

I In OCCI viewI One compute resourceI One sensor resourceI One collector link

Cloud Forward2015

Augusto Ciuffoletti

The web server

I The web server is deployed (with OCCI docs)I this is part of the cloud infrastructure

Cloud Forward2015

Augusto Ciuffoletti

The web server

I The web server is deployed (with OCCI docs)I this is part of the cloud infrastructure

Cloud Forward2015

Augusto Ciuffoletti

The dashboard

I The dashboard is deployed (UDP to receive data)I this is on user premises

Cloud Forward2015

Augusto Ciuffoletti

The dashboard

I The dashboard is deployed (UDP to receive data)I this is on user premises

Cloud Forward2015

Augusto Ciuffoletti

The computing resource

I The virtual computing resource is deployedI the RMI interface is for configuration

Cloud Forward2015

Augusto Ciuffoletti

The computing resource

I The virtual computing resource is deployedI the RMI interface is for configuration

Cloud Forward2015

Augusto Ciuffoletti

The sensor

I It is the resource that implements the monitoringI uploads the collector across the RMI interface

Cloud Forward2015

Augusto Ciuffoletti

The sensor

I It is the resource that implements the monitoringI uploads the collector across the RMI interface

Cloud Forward2015

Augusto Ciuffoletti

GO!

I The dashboard receives the probe measurements...I ...filtered by the sensor

Cloud Forward2015

Augusto Ciuffoletti

GO!

I The dashboard receives the probe measurements...I ...filtered by the sensor

Cloud Forward2015

Augusto Ciuffoletti

Conclusions

I Microservices are appropriate for distributedmonitoring

I The OCCI API is appropriate as the interfaceI A prototype is available on dockerhub (look for

occimon-live)I BYOD demo on stage tonight at 17:45

Cloud Forward2015

Augusto Ciuffoletti

Conclusions

I Microservices are appropriate for distributedmonitoring

I The OCCI API is appropriate as the interfaceI A prototype is available on dockerhub (look for

occimon-live)I BYOD demo on stage tonight at 17:45

Cloud Forward2015

Augusto Ciuffoletti

Conclusions

I Microservices are appropriate for distributedmonitoring

I The OCCI API is appropriate as the interfaceI A prototype is available on dockerhub (look for

occimon-live)I BYOD demo on stage tonight at 17:45

Cloud Forward2015

Augusto Ciuffoletti

Conclusions

I Microservices are appropriate for distributedmonitoring

I The OCCI API is appropriate as the interfaceI A prototype is available on dockerhub (look for

occimon-live)I BYOD demo on stage tonight at 17:45