I have a couple of cubes that I'm wanting to designate for real-time ROLAP storage, so I've set up the cubes, and each of their dimensions, to use Proactive Caching. The cubes are defined to use a Scheduled Polling query notification scheme, and the dimensions are set to use SQL Server notification.
According to perfmon (the "Total Deadlocks Detected" counter), this setup is somehow resulting in numerous deadlocks. I tried disabling proactive caching for all but two dimensions, but still encountered this problem. Has anyone else encountered this, or have any information as to why this configuration is resulting in deadlocks?
The server is running SQL Server 2005, SP2. My workstation (on which I can reproduce this issue) also has a dual core processor.
Thanks,
Joe Miller