Fell Into SCOM Role Operations DB using 700MB disk IO?

[npwiley]

So I am brand new to SCOM, basically a guy left who managed it and now on of the vms (Operation DB) is eating up 700MB+/s of disk IO constantly. Any suggestion on where I might look at what is causing this? It has a local SQL server installation and sqlservr is the process eating up all of the IO. Talking with a database team member they tell me OperationsManagerDW.dbo.StandardDatasetMaintenance is what is eating up the db. Any help would be GREATLY appreciated.

 

Thanks!

[AdinE]

Here's the link to the SCOM Sizing Helper tool: https://blogs.technet.microsoft.com/momteam/2012/04/01/operations-manager-2012-sizing-helper-tool/

 

This is the tool we use when we are planning and designing the architecture of SCOM. However, if you plug in some info about your environment, you might be able to see what the predicted load is, and compare.

 

Further, for the SCOM databases, it is best practice to have them installed on a dedicated SQL Server. If your environment has a SQL cluster, then a dedicated SQL Instance. Normally we don't have SCOM's DBs running from the Management Server, unless it's in a POC, LAB, or DEV environment.

 

Can you DBAs comment to the installation/configuration of SQL? Is there sufficient RAM on the VM? If not, maybe it's paging out to disk?

[npwiley]

The SCOM setup is completely virtual, Operations DB is on one server, DataWarehouse is on another, Report server/web console another. It looks to be monitoring about 600 clients. The weird thing is that on the server with the high IO it is just doing 700MB/s or reads on SQL Server with very minimal network traffic, so it does not appear to be reading data for another machine. I will look into over design of the environment at some point hopefully, but for now im just trying to put out this fire so it can continue to run.

 

Its equipped with 22GB of RAM, which it is close to maxing out, but it does very little reads/writes to the pagefile.

 

Thanks for your help so far anything else you would suggest?

[AdinE]

Based on your original comments, if the issue is coming from OperationsManagerDW.dbo.StandardDatasetMaintenance, then maybe the grooming is set too aggressively.

 

In this thread (https://social.technet.microsoft.com/Forums/systemcenter/en-US/dee32b4e-8cca-4661-83fb-fcb40a9536bf/cpu-usage-100-on-sql-opsmgr-db-and-dw-server?forum=operationsmanagergeneral) someone mentioned issues with certain update to SQL Server (what version/update is your SQL at)? Also in that thread, there was a comment about potential issues with HDD firmware.

 

Here's another article on grooming: http://blogs.technet.com/b/sudheesn/archive/2010/08/03/gooming-in-scom.aspx