Page 509 - Cloud computing: From paradigm to operation
P. 509

Cloud Computing management                                          2


            6.3     Cloud service lifecycle management stages

            A cloud service is managed differently in the various stages of its lifecycle. At a high level view, the lifecycle
            of the cloud service includes four stages: design, deployment, operation and retirement. The term stage is
            used to designate the various steps in this overall process. Particular tasks in each stage and how to control
            the level of independence or interaction from one stage to another stage vary from one organization to
            another.

            The tasks in each stage depend on development process considerations (e.g., software development, testing,
            deployment, etc.) and operational or business considerations (e.g., how a software component becomes a
            service in the organization and what are the resulting constraints that are then imposed on the software).
            Such considerations are part of CSP operational style. This clause presents the basic stages description below
            but in reality, the involved roles and implementation will vary based on the actual conditions.
            The cloud service lifecycle management stages are as follows:

            1)      Stage 1: Design
            The service design metadata is built or the existing one is modified by creating new versions or variants in
            order to capture at the service level the necessary artefacts to support the business considerations.
            During the design stage, the necessary service process specification, template, rules, etc. are created and
            developed for the operation stage. Policies and their enforcement points are also defined for different service
            conditions.
            2)      Stage 2: Deployment

            The more specific service deployment metadata specialized from the same service design metadata is built.
            It  is  typically  done  by  associating  specific  values  or  value  ranges  for  each  invariant  non-functional
            characteristic.  Several  sets  of  service  deployment  metadata  associated  with  the  same  service  design
            metadata  constitute  in  a sense  a specific  family of related  service  deployment  metadata. These  service
            deployment metadata sets will be mapped to different product offerings and will be used as essential entry
            points during the fulfilment process.

            3)      Stage 3: Operation
            This stage covers all the activities related to the actual instantiation, monitoring, analysis and feedback to
            design stage of each service. The management tasks of this stage actually involve two aspects:
                    a)  the cloud service itself (seen as a coordinated set of resources in the execution environment
                        from one or several domains);
                    b)  the cloud service operation metadata, which is the unique representation of the service in the
                        management infrastructure of the CSP.
            The  operation  stage  includes  the  service  instantiation  and  delivery  process,  during  which,  the  designed
            templates and policies are distributed to the involved roles. The service is instantiated based on CSC's request
            and  CSP's  infrastructure  condition.  The  monitoring  control  mechanism  is  set  up  along  with  the  service
            instantiation.
            The automation of configuration is also provided in this stage, including the initial and subsequent changes.
            To achieve automation, the inventory, monitoring and control functions are activated.

            During the whole operation stage, the monitoring activities are performed via event listening; computing
            analysis is enforced based on data collection. The event, requiring healing and/or scaling, is published, based
            on  pre-defined  policies.  The  actions  (healing  and/or  scaling)  are  preformed  to  implement  the  changed
            demands.
            Based on the analytical data obtained during the operation stage, the patterns governing usage, thresholds,
            events, policy effectiveness, etc., are discerned and the feedback necessary to effect changes in the design
            stage is enabled.




                                                                                                         501
   504   505   506   507   508   509   510   511   512   513   514