Multi-Level SLA Management Architecture for Cloud Computing
The SLA for a service must be based on realistic, achievable targets (e.g. for performance and availability), and the achievement of these targets depends on the performance of the internal and external services that underpin the delivery of the main service. Putting it another way, SLAs must reflect the levels of service actually being delivered or that can be delivered. They are about what can be done rather than what the authors would like to be done. If a customer requires a different level of service, this would normally be dealt with by raising a service level requirement.