The key aspects of an operational system are:
A service is said to be available if it is capable of performing the task it is designed to perform at the appropriate time. This is a "yes-no" concept, because the service is either available or unavailable.
When considering unplanned downtime, this "yes-no" concept is based on a failure model called "crash failure". However, real system failures are usually more complex.
This is measured by the ability to meet certain pre-defined criteria such as throughput of the system (for example, in number of users supported) and average response time for each user. Performance is said to be acceptable when a certain level has been achieved over a given period.
Performance clearly depends on availability. However, availability does not always guarantee performance. For example, an SAP system where the CPU utilization of the database host is very high has low performance, but it could be considered as an available system. In practice, the distinction between the two aspects often blurs in extreme cases (that is, extremely poor performance means the system is considered to be unavailable or "down"). For instance, a network connection might be terminated (that is, be unavailable) after a time-out has occurred due to unacceptably slow performance. Even though this documentation focuses on the availability aspects of the system, the close relationship to the performance aspects should not be ignored.
To increase the availability of a system, it is essential to minimize downtime. Downtime can be categorized as planned and unplanned, and at a more detailed level, it can be separated into planned and unplanned downtimes for the various services which make up SAP systems.