Take the following actions before implementing workload management.
|Monitor single system metrics and resource usage trends within specified time frames with the Viewpoint Metric Heatmap portlet.||For information on the Metric Heatmap portlet, see Teradata® Viewpoint User Guide, B035-2206.|
|Study resource contention at various times of day and to understand blocking and wait time patterns with the Viewpoint Query Monitor and Lock Viewer portlets.||For more information on the Query Monitor and Lock Viewer portlets, see Teradata® Viewpoint User Guide, B035-2206.|
|Enable DBQL detailed logging for at least 3 weeks to understand request performance.
|Enable ResUsage.||ResSPMA, ResSPS, and ResSAWT are extremely helpful with system, scheduler, and AMP Worker Task (AWT) analysis.
The ResUsageSPS table can help you determine resource usage by workload.
|Identify workload types, such as ad hoc, ETL, and short highly-tuned requests:
|Document all Service Level Goals.|
|Document the exceptional conditions that now require the DBA to intervene, such as long request runtime, high request CPU use, high skew, and too many sessions or requests.|
|Document current operational limits, such as optimal concurrency levels for load utility jobs and all application groups.|
|Ensure the Viewpoint server software is current.|
Everyone involved with the Teradata system needs to understand and agree about business priorities. Many workload management performance issues are caused by user disagreements about the relative importance of the workloads.