Monday 6 October 2014

Business Monitor and Cognos performance problem


Recently we have been seeing some issues with our Business Monitor servers on start up.   After running IBM Business Monitor 7.5.1 for over a year, every time we restarted the environment there was a spike in CPU consumption that lasted for over half an hour.

On closer inspection it was the cognos process spawned by Monitor that was consuming all this CPU.    A quick call to IBM support pointed us at the NC tables used by cognos.  A few of our tables had over half a million rows, where as the following document suggestes anything over 1000 could impact performance:

http://www-01.ibm.com/support/docview.wss?uid=swg21637944

So we followed the steps in the technote and cleared down the NC tables and all of a sudden there was no issue following the next restart.

We have been keeping an eye on these tables since then and noticed they do continue to grow.   Further questions to IBM support suggest this is normal behavior as long as there is no rapid increase in the size of the tables.  It just means every now and then we may have to clear down these tables again.