June 11, 2019 at 3:25 pm
I haven't been able to find an answer for this via Google. Hopefully the forum can help.
We recently upgraded a SQL 2016 SP2 instance from CU3 to CU6 and we're seeing more TempDB contention. Does anyone know of a change in CU6 that would explain different behavior in TempDB?
Thanks in advance.
June 11, 2019 at 7:42 pm
Haven't made it to CU6, yet (we try to stay off the bleeding edge) so thanks for the heads up. Hopefully, this will also bump your post back close to the top because this seems to be a wee-bit important.
--Jeff Moden
Change is inevitable... Change for the better is not.
June 11, 2019 at 9:46 pm
Define contention. Are you seeing lot of pfs/sgam wait types? Are any on your custom temdb settings still there post applying CU6?
June 12, 2019 at 8:06 am
and how many files you got on tempdb?
and can you go to CU7 just in case it fixes the issue?
June 12, 2019 at 2:32 pm
Yes, we are seeing sporadic pfs/sgam wait types. We currently have 40 TempDB data files. We are thinking about going up to 48.
June 12, 2019 at 2:56 pm
How many logical and physical cores do you have? We have a 48 physical core and 96 logical cores box, we found 48 files for us seems to be helpful. Be cautious about adding files, adding too many files can also cause issues. Sporadic pgs/sgam is normal, in past i sampled wait types every 5 secs and got an aggregate total on tempdb waits, when you add more files your goal should be to reduce that number.
June 12, 2019 at 3:14 pm
We have 96 physical cores and 192 logical processors.
June 12, 2019 at 3:20 pm
You can start with 48 and keep adding 4 files until you see acceptable improvement.
Viewing 8 posts - 1 through 7 (of 7 total)
You must be logged in to reply to this topic. Login to reply