September 8, 2016 at 2:27 pm
I am setting up new SQL2016 Failover Clusters and I have noticed something that concerns me. I am working with Identical Setup Windows Server 2012R2 and SQL2016/SQL2012 with latest SP/CU on the same Virtual hosts and Identical network and disk setup. When I run TPC-C benchmarks against SQL 2016 I get about 1/6 the performance versus SQL2012 or SQL2014. I have tried compatability modes and changing the Cardinality Estimator to legacy mode. Does anyone have any idea what I might be overlooking? Has anyone else seen similar results?
September 10, 2016 at 1:30 pm
Which tool are you use for performance testing? HammerDb?
September 10, 2016 at 2:10 pm
I used the latest version of HammerDB.
September 11, 2016 at 1:22 pm
If it is test envoirment I think firstly
DBCC DROPCLEANBUFFERS
and
DBCC SQLPERF('sys.dm_os_wait_stats', CLEAR)
and start the HAMMERDB after the test you can
look at the wait statistics it can give some tip about the problem
maybe it can be related to MAXDOP ,tempdb number etc...
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply