The default connection pool parameters set when JDBC data sources are created through the IBM FileNet Configuration Manager are generally sufficient for standard workloads but can be a bottleneck during high volume conditions or long duration transactions. Also, the new IBM FileNet Content Platform Engine’s database connection feature which allows sharing of a single pair of JDBC data sources among multiple object store and isolated regions can result in higher number of active connections.
Below is an example of a JDBC Data Source created by the IBM FileNet Configuratoin Manager tool in IBM WebSphere Application Server 8.5. The default value for Maximum connections is 10.
The Tivoli Performance Viewer which is included in WebSphere Application Server gives you an easy way to monitor the JDBC Data Sources connection pool. Below is a screen shot of the FNOSDS and FNOSDSXA JDBC Data Sources for one of my object stores as viewed through the Tivoli Performance Viewer right after the Content Engine has been started. Note the PoolSize for the non-XA data source is 7 and for the XA data source is 0.
After adding several thousand documents to this object store, I refreshed the page in the Tivoli Performance Viewer. The PoolSize setting for the non-XA data source hasn’t changed, but for the XA data source is now 3.
After reviewing the values in the Tivoli Performance Viewer, with the observed workload, I can conclude the non-XA and XA data sources connection pool’s ‘maximum connections=10’ value is sufficient for this object store.
If the PoolSize values for either JDBC Data Source had been 10 then I would have concluded the ‘maximum connections’ value in the JDBC Data source is set too low and there is a potential performance bottleneck. Increase the values for both JDBC Data Sources and repeat the above test. There is a similar statistic monitoring functionality in Weblogic for monitoring JDBC Data Source connections.
I hope that the above information sheds some light on how to determine the most efficient connection settings for an Object Store’s JDBC data sources with the IBM FileNet Content Engine application.