19.3. Availability-Level Server Classifications

The computi0ng industry has defined four availability-level (AL) classifications for servers. These classifications define the consequences of a system failure, and are described in the following table.

Level	Characteristics
AL1 Data availability	Work stops. An uncontrolled shutdown results. Data integrity is ensured. System is available 95% of the time.
AL2 Standard availability	User is interrupted, but can log on again. User might have to rerun some transactions from a journal file. Possible performance degradation can result. System is available 95% to 99% of the time.
AL3 High availability	User stays online. Current transaction might need to be restarted. Performance degradation can possibly result. System is available approximately 99.9% to 99.99% of the time.
AL4 Continuous availability	Process is transparent to user. No work is interrupted. No transactions are lost. No performance degradation results. System is available 99.999% to 100% of the time.

Standalone ProLiant servers are rated AL1. If the server is down, all work being performed by that server stops. ProLiant clusters can achieve AL2 or AL3. The user might notice a small interruption or degraded performance. HP also provides systems ranked AL4, including the NonStop Himalaya systems. Guaranteed uptime requires HP Global Services.

Although every business would like its servers to be always available, many businesses do not operate 24 hours a day, so availability greater than AL1 is not driven by the operating requirements. However, to maximize server availability within that classification, a company can employ technologies and create configurations that minimize time lost as a result of problems, including the following:

• Increasing availability through power protection (discussed in this chapter).

• Increasing memory availability to reduce memory errors and faults (discussed in this chapter).

• Optimizing storage for fault tolerance and recovery (discussed in the next chapter).