Technical Blog Post
Abstract
Message Broker agent: CPU usage
Body
We don't have formal performance figures so all information given below are based on experience of support teams.
There are many factors that may affect CPU usage.
Generally speaking, it depends on how many EG/MF are deployed in the broker and if message broker agent (KQI) is configured to get detailed data.
1. In kqi.xml, set defaultCollectNodeData or collectNodeData to "YES".
It will show all nodes data in the workspace 'broker status'.
It might cause CPU or memory issue if you have more than 1000 nodes in one broker.
2. Enable Archive Statistics data with detailed node level.
There are two related parameters for the command 'mqsichangeflowstats'
'-t threadData' specifies the thread statistics level: values allowed are 'none' or 'basic'.
'-n nodeData' specifies the node statistics level: values allowed are 'none', 'basic' or 'advanced'.
Setting detailed node level data (specifically, set 'advanced' for nodeData) will extremely increase the data parsed by KQI agent if you have lots of nodes.
3. If you have hundreds of message flows deployed and enable archive statistics, it will also cause performance issue even if you don't set detailed node level.
4. If you enable historical data but do not configure Data Warehouse connection, the local historical file can increase indefinitely to more than 1GB. Reading and writing this huge file will cause high CPU issue.
5. Number of transactions on broker side do NOT have impact on agent performance.
Subscribe and follow us for all the latest information directly on your social feeds:
|
|
|
Check out all our other posts and updates: | |
Academy Blogs: | https://goo.gl/U7cYYY |
Academy Videos: | https://goo.gl/TLfMoF |
Academy Google+: | https://goo.gl/HnTs0w |
Academy Twitter : | https://goo.gl/AhR8CL |
UID
ibm11084755