I was helping troubleshoot SCOM performance issues with a customer yesterday and came across something which could be contributing to the performance problems they’re experiencing.
They’re monitoring a large SQL estate, including a lot of SQL clusters (around 800). It seems that the SQL MP (2014 , but possibly also others) is creating a (sub)-group for each of those clustered pairs (ie. around 800 Groups, just for the SQL clusters) .
Is anyone familiar with this and, if so, have they seen it cause any performance problems in their or other environments? Thanks
It does indeed create a lot of groups, but i cannot see we have sub groups for each and every SQL Cluster. We do have always on availability group objects within the SQL Server Always On Availability Group, but no sub groups.
Altough, Windows Clusters does have individual group objects, could that be it? Also, dynamic group calculation can be performance impacting when you have a large number of groups/complex querys. Originally calculation run often (every 60sec or so) we have set this to 15 minutes.
Thanks Ehrnst, don’t know if this is the answer but you could be onto something with the dynamic group calculation, I think this is something that they’re doing across all groups, so definitely worth putting this to the customer, I’ll flag it for them and provide an update.
It would be interesting to see if there is any monitoring targeted at the cluster groups. If there is none or none that is worthwhile, perhaps disabling the discovery that creates the groups could be justified.
We too have increased the Group Calculation Interval to 900 seconds and would recommend following the link from @Ehrnst to get Kevin’s feedback as well.