An ongong check of various statistics and events, which can be used to spot telltale signs of underlying issues.

Database management systems can be monitored to keep an ongoing check of various statistics such as memory usage, excessive time spent in certain operations such as table scans, excessively high cache miss statistics or excessively long I/O wait times. Events such as deadlock resolution (this involves killing one of the deadlocked processes), timeouts or warnings raises on exceptional conditions can also have significant diagnostic value.

This information can be retrieved through views in the system data dictionary or event trace data exported by the system. Monitoring tools that use this data may be provided by the vendor or third parties or built on an ad-hoc basis.

history | excerpt history