If the performance of Adaptive Server at the server level is not acceptable, and you are running 15.0.2 ESD #3 or later, you may shut down the Adaptive Server and restart it with trace flag 757 set in the RUN_server file. This is also effective when you experience unusually high levels of CPU usage without a clear cause, while running a multiengine Adaptive Server.
If the procedure cache is filled with idle cached plans, and the CPU usage is not high, run the following dbcc commands instead. However, using these commands is likely to have a lesser effect than restarting the server.
dbcc traceon(757) go dbcc proc_cache(free_unused) go
Do not use trace flag 757 in Adaptive Server versions
earlier than 15.0.2 ESD #3.