Page 86 - Cloud computing: From paradigm to operation

1 Framework and requirements for cloud computing

It is important to distinguish between maintenance performed by the cloud service provider and
maintenance performed by the cloud service customer. In the case of an SaaS service, it is likely that virtually
all maintenance actions will be performed by the provider. In the case of IaaS and PaaS services, the
application components belong to the cloud service customer and the cloud service customer is responsible
for the maintenance of those components. The provider is responsible for the environment in which the
application components run, which varies depending on the details of the service, but which might include
such elements as the hardware resources, operating system or middleware.

On the one hand, it can be in the customer's interests that a service or service platform be upgraded or fixed.
On the other hand, any changes to the behaviour of a service can have a negative impact on the customer,
possibly requiring changes to application components and to customer ICT systems or calling for retraining
of customer service users. As a result, it is important that maintenance of services is subject to governance
practices that are transparent to the customer.
Maintenance practices should be documented in the SLA for the cloud services and should include the
capability for the customer to report problems and request fixes and also a mechanism for the cloud service
provider to notify the customer of pending maintenance changes and their schedule.

Versioning is the appropriate labelling of a service (or of components of a service, such as the operating
system level used in an IaaS service), so that it is clear to the customer that a particular version is in use. It is
important that the service be given a new version label when maintenance of a cloud service occurs.

Where significant changes are made to a service between two versions, the older version of the service
should be available in parallel with the new versions for an agreed period of time.

8.5.7 Performance
Performance includes a set of non-functional facets relating to the operation of a cloud service such as:

• availability of the service;
• response time to complete service requests;
• transaction rate at which service requests are executed;
• latency for service requests;
• data throughput rate (input and output);
• number of concurrent service requests (scalability);

• capacity of data storage;
• (for IaaS and PaaS) the number of concurrent execution threads available to an application;
• (for IaaS and PaaS) the amount of memory (RAM) available to the running program;
• data centre network IP address pool and/or VLAN range capacity.

Where the service involves running an application (IaaS, PaaS), the same facets of performance apply to the
behaviour of the application running in the cloud service provider's environment.

Depending on the charging model, the ability of the cloud service to scale its use of resources in accordance
with the terms of the SLA can also be an important facet of performance. Performance should have metrics
defined in the SLA for each performance condition identified and these metrics should be monitored during
operation of the cloud service to ensure that the service meets the performance terms of the SLA.
8.5.8 Portability

Portability is significant in cloud computing since prospective cloud service customers are interested in
avoiding lock-in when they choose to use cloud services. Cloud service customers need to know that they
can move cloud service customer data or their applications between multiple cloud service providers at low
cost and with minimal disruption. The amount of cost and disruption that is acceptable can vary based upon
the type of cloud service that is being used.

For example if a cloud service customer organization is considering moving from one IaaS cloud service
provider to another, the cloud service customer should be able to take its data and the virtual machine (VM)

78