The workload event CPU Utilization helps a DBA identify and solve issues before they worsen. The CPU Utilization event action can notify a DBA to solve the underlying issue before it causes unacceptable performance.
The CPU Utilization event can detect and resolve the following situations:
- A workload has taken over the system: If workload Godzilla has CPU Utilization > 90%, change the health condition to Blue, which lowers the throttle limit on workload Godzilla. Send the DBA an alert so he can investigate further.
- A workload is starved: Workload Hungry has an event combination for CPU Utilization and Arrivals. If CPU Utilization < 2% and Hungry has Arrivals > 100 per hour, send an alert to the DBA to investigate why Hungry is being starved.
- As an alternative to a user-defined event to activate a planned environment: If Night_Owl workload has CPU > 0% for 10 minutes, set the health condition to Loading, which lowers the throttle limits on non-load requests.
- High workload CPU utilization: If workload Popular has CPU > 50% for 30 minutes, write an event notification to DBC.SystemQTbl to notify Popular application users not to submit new work. It is assumed an application outside TASM/TIWM monitors DBC.SystemQTbl rows for this condition and then sends user notifications once this condition is detected.
Teradata recommends using the Averaging qualification option for this event to distinguish between a temporary CPU Utilization level and a more chronic one.