We recently experienced a significant performance degradation on our main TP SQL server (Current primary of SQL2012 - 3 servers in High Availability config)

This was diagnosed to the OBJECTSTORE_LOCK_MANAGER consuming approx. 66% of available memory (40gb) which it would not release.

Running 

cleared the problem

Investigating log files has not provided info to determine how this situation occurred. What may have caused OBJECTSTORE_LOCK_MANAGER to reach this state?

You could have given DBCC MEMORYSTATUS dump for analysis but since you cleared buffer cache its of no use we have to wait until you face the issue agin.

This clerk tracks down memory utilized by locks. Each lock which is taken has memory associated with it. More memory taken by locks signify SQL server is using too many locks and is not able to go for lock escalation have you enabled trace flags 1211 and 1224 which Might prevent lock escalation. It also signifies lock is being taken not released, do you see deeadlock and blocking ?

For excess and long time blocking please see

http://support.microsoft.com/kb/224453/en-gb

I would say a poor query design can cause this. Of course to comment more I need more details.

Check this similar post ..

http://social.msdn.microsoft.com/forums/sqlserver/en-US/f7fcc0f8-953f-4aff-89fd-3e80e34e3255/high-objectstorelockmanager-what-that-mean

Thanks for the info

We did not have these trace flags enabled.

Further analysis shows that we had an event a few days previously which probably caused this condition (although we cannot tell exactly what it was)