Cardinality Statistiscs consuming max CPU resources
Article ID: KB0083014
Updated On:
| Products | Versions |
|---|---|
| TIBCO Data Virtualization | - |
Description
The four (4) server nodes on the cluster are experiencing high CPU utilization when a Statistics Gathering is kick-off by CIS.
Issue/Introduction
Cardinality Statistiscs consumin max CPU resources
Resolution
Resolution
Validated that the issue was happening while stats gathering and the cardinality stats parameters need to be optimized.
There is a field within Statistics Tab for a given datasource where you can configure the number of threads per each datasource.
Recommendations provided:
•I see that you have already configured 10 threads. Would it be possible to drop it to 5 or even 1 and try this weekend. You are initiating many threads from every datasource and it could be a possible reason for the spike.
•At the same time collect column boundary stats instead of running full statistics. Please remember that this will change the execution plan. Column boundary gets only the min/max/distinct of the columns, but does not build histogram as in the case of full stats.
Validated that the issue was happening while stats gathering and the cardinality stats parameters need to be optimized.
There is a field within Statistics Tab for a given datasource where you can configure the number of threads per each datasource.
Recommendations provided:
•I see that you have already configured 10 threads. Would it be possible to drop it to 5 or even 1 and try this weekend. You are initiating many threads from every datasource and it could be a possible reason for the spike.
•At the same time collect column boundary stats instead of running full statistics. Please remember that this will change the execution plan. Column boundary gets only the min/max/distinct of the columns, but does not build histogram as in the case of full stats.
Feedback
Yes
No
Powered by
