17.10 - Delay Queue Depth and Delay Queue Time - Advanced SQL Engine - Teradata Workload Management

Teradata Vantageā„¢ - Workload Management User Guide

Product
Advanced SQL Engine
Teradata Workload Management
Release Number
17.10
Release Date
July 2021
Content Type
Administration
User Guide
Publication ID
B035-1197-171K
Language
English (United States)
Two workload events can detect when a workload is delayed for a long time because of throttle limits:
  • Delay Queue Time: This event detects how long in seconds, minutes, or hours the longest-waiting request has been in the delay queue. Delay Queue Time can, optionally, include the time a system throttle delays a request.
  • Delay Queue Depth: This event detects how many requests are currently delayed.

An interesting use case follows: Some applications indicate performance levels to users before they submit their requests. This levels system load because users may defer their next request when there are long response times. But even if the users submit their requests anyway, they are more satisfied because they know what to expect. In fact, one customer saw fewer users killing and resubmitting their requests when the customer established user expectations with a response time indicator. Using the response time indicator improved overall system performance because it eliminated the kill and resubmit resource consumption. You can use delay queue events to serve as an application performance indicator when you select the Post to QTable action. This action writes rows to DBC.SystemQTbl. Monitoring applications consume the rows in this table to provide the performance level indicators, as described here.