2.6 Managing Service Levels with Oracle Database QoS Management

The implementation of Oracle Database QoS Management is completed by actively managing the service levels, which means responding to alerts, reviewing and implementing recommendations, and tracking results. This section describes the actions you would perform on the demo system.

After all the workloads run and the Dashboard displays the performance of the demo system, you need to be alerted should a workload surge or failure cause a Performance Objective to stop being met. The Performance Satisfaction Metric (PSM) normalizes all of the objectives and provides a quick way to observe the health of the system. By observing the PSM Trend indicator you can see how well a Performance Class is meeting its objective over the last five minutes, and problems can be observed. Performance Objective violations produce recommendations that state how resources should be reallocated to relieve the bottleneck. Details and projections are available for further analysis of the bottleneck and possible solutions. If the recommendation is an action that can be implemented by Oracle Database QoS Management, then an Implement button is displayed.

Performance Objective violations of short duration are tolerated in most SLAs. Therefore, Enterprise Manager alerts can be configured by Performance Class specifying the duration of continuous violation. These alerts are configured on the Database alert page, but can be defined for all Performance Classes in the cluster.

An audit log of policy changes, violations and actions is available in the Oracle Grid Infrastructure home in the $ORACLE_BASE/crsdata/host/qos/logs/dbwlm/auditing directory on the server that hosts the Oracle Database QoS Management server. To determine which server is hosting the Oracle Database QoS Management server, enter the following command at the operating system prompt:

srvctl status qosmserver