Microsoft SQL Server 2000 Administrators Companion

3 4

As a database administrator, your primary job is to keep the database up and running during specific time periods, which are usually outlined in a service level agreement. This service level agreement probably specifies the amount of uptime your system must provide, as well as performance rates and recovery time in the event of a failure. Using MSCS can increase the amount of uptime and decrease recovery time. Although server hardware, Windows 2000, Windows NT, and SQL Server are usually stable and reliable, components sometimes fail. In fact, a variety of types of failures can occur in a complex computer system, including the following:

• Disk drive failure Disk drive technology has improved, but a disk drive is still a mechanical device and, as such, is subject to wear. The disk drive is one of the most common areas of failure.
• Hardware component failure Hardware failures can occur because of wear and tear on the components, primarily from heat. Even the best-made computer equipment can fail over time.
• Software component failure Some software flaws are discovered only under rare conditions. Your system might run for months or years until a specific set of conditions uncovers a problem. In addition, adding applications to a stable environment might modify a critical library or file and cause problems.
• External failure A system can fail because of external causes, such as power outages. Whether your system can survive such a failure depends on whether you are using an uninterruptible power supply (UPS) and redundant power sources.
• Human error Clustering does not usually protect a system against failures caused by human error (such as accidentally deleting a table or a Windows NT file system partition).

Failures are unavoidable. The issue of how to best prepare for some of these failures will be our focus in this chapter.