The servers have 512GB of memory with sql allowed to use 348GB. The numerous data files are all on a Netapp SAN device. The main user database is 3.3TB with an aspstate database also receiving a lot of traffic. We have used a large number of LUNs to spread out the data files, with the transaction log for the main database on its own lun and the four tempdb data files and log file all on their own Luns. Going to add four more same-sized tempdb data files this sunday.

Most of the problem queries come from the application ( ORM generated ) and having reporting and OLTP traffic all on one database is a problem --- to be addressed along with a plan to archive/partition millions of older records. Sql agent jobs and batch jobs run by the application scheduler are spread out as much as possible.

Since we have NEVER seen database corruption from these "15-second" messages, it is mainly something upper management gets in a tizzy about. 15 seconds? Nothing is taking 15 seconds.

Sql server 2012 Enterprise 64-bit on Servers 2008R2 64-bit ( prod databases still in 2005 compatibility mode )