6.2 Guidelines for Exadata Monitoring

Regardless of the information source, it is considered good practice to maintain baseline collections of the various performance data sources covering the important periods of your workload, such as peak periods, month-end or year-end processing, or before patching or upgrades.

Baseline information enables a comparison with a known 'normal' state when a performance regression occurs. Such comparison helps to identify changes and guide further analysis.

Baseline data should include AWR data (using AWR exports or AWR baselines), Exadata metrics, and ExaWatcher collections. For data warehousing workloads that rely on smart scans, keeping a collection of SQL monitor reports for key SQL statements is also highly recommended.