161
After the Cloud service has been implemented and has assumed productive
operation, the following topics become relevant:
Support: The use of Cloud services primarily means a change in
service processes. In this regard, it must be defined who will be
responsible to document and communicate, as well as possibly scale
up the service requirements, and how this is to be done. The
different support possibilities will have already been clarified during
the process of selecting a Cloud service and Cloud provider, and
must be communicated to the users.
Performance: Potential bottlenecks should be proactively
communicated by the Cloud provider on the one hand, and, if
necessary, be double checked by internal performance
measurements. In this respect, according to the defined service level
agreements (SLAs), the person responsible for the service is required
to verify compliance with the performance and availability
guarantees as well as to scale to more capacity as necessary.
Availability: In a continuous cycle, the service provider will
implement patches and updates. The respective information must be
made available by the service provider and users of the service must
synchronise it with their own update cycles.
Information on compliance with the SLAs and parameters is generally
delivered by the Cloud provider. The company itself is responsible for
analysis and monitoring. The respective measures must be coordinated
between the company and Cloud providers and should be intended to
achieve compliance with the SLAs agreed upon.
Identifying performance indicators
The indicators and parameters necessary for monitoring Cloud services
should already have been defined as "IT quantities" in the preparation
process. They can also be identified by a careful review of the contracts with
the Cloud provider. These parameters are defined and documented in
service level agreements (SLAs).
Typical metrics for Cloud service performance are:
Availability in % of the time that the service is in operation.
Guaranteed up‐and download bandwidth.
Response times in the event of failure.
