February 18, 2015 at 3:14 pm
Hello,
I'm diagnosing large performance variations during our automated nightly maintenance over the past few months. It seems to be related to parallelism, TempDB usage & disk performance during index rebuilds, or perhaps an impending RAID disk failure. The problem is isolated to our production environment. We'd like to avoid the cost and effort of duplicating the full production environment in a sandbox, instead focusing on migration to a new production system.
My server's MAXDOP is 0 - the recommended setting for an 8-core server.
The Cost Threshold for Parallelism is the default value of 5. I'd like to adjust this incrementally to track performance impact.
Is it considered safe to tweak the Cost Threshold for Parallelism on production systems, or would it be wiser to duplicate the production environment and trial it there?
Eric
February 19, 2015 at 10:34 am
Suggest to do it on Non-Production environment first if you have.
February 23, 2015 at 7:22 am
Thanks for the input, CoolDBA.
Eric
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply