This white paper is a guide for DBAs about determining the proper thresholds for monitoring a SQL Server environment.
Monitoring SQL Server is a critical component of a Database Administrator’s, or DBA's, job. The bigger the environment is, the more important monitoring becomes. Enterprise environments typically have a high ratio of servers to DBAs and no DBA can keep watch of every aspect of the environment on their own. A very important aspect of monitoring is alerting.
Monitoring alone isn’t enough. When certain resource metrics or counters are reached, the DBA needs to be alerted and, in some cases, immediately notified that the threshold has been reached. Alerts can be delivered in a variety of ways and should not be confused with notifications. Alerts can be written to a log, raised in an application interface, or sent to another system for processing. A notification is a response to an alert that informs a person, or a group of people, that a specific alert has been raised.
Monitoring, alerts, and notifications all work together to form a good monitoring solution. Any good monitoring tool will perform these three basic functions. An additional function that is not part of all monitoring tools is baselining. Baselining is important for effective monitoring as most metrics or counters can vary greatly from system to system.
There are only a handful of counters that can unmistakably state that a specific value indicates a problem. Most counters have to be watched over time to establish an acceptable threshold for a healthy system. A baseline is a continual process. You can’t just baseline once. As systems evolve, so will their performance profiles. So a really good monitoring tool will have a built-in baselining functionality.
Register to read the full whitepaper.