1 Reply Latest reply on Feb 10, 2015 6:47 PM by mandevil

    Why are all tables stored in the PRIMARY filegroup?

    mdean@cvent.com

      While troubleshooting performance issues related to the DPA application, I noticed that all tables are stored in 1 file within the PRIMARY filegroup. I see that tables are created on-the-fly for the corresponding days of the month. Is there a way to configure the application to spread out the IO? (Besides creating another filegroup and setting that to the default FG. I don't want all tables in 1 FG.)

       

      BTW, the performance issues we're seeing is happening every 10 minutes. I'm assuming this is when DPA aggregates the data to show in the previous 10-min time slice.

        • Re: Why are all tables stored in the PRIMARY filegroup?
          mandevil

          How is your storage configured/architected?  Are you hitting storage on a SAN?  Local disk?

          As far as the objects spinning up every month, you may be referring to our monitored instances tables.  We create a new set of tables for every monitored instance.  That's what the _id represents and helps us map back to the COND table.  I'm not aware of objects being created monthly, but if you give me an example, I can certainly look into it.

           

          Every 10 minutes we do a couple of things:  We do our stats poll that can potentially pull back a lot of data that needs to be persisted to the repo.  We also do metric summarization which results in reads/writes.  If you are seeing performance issues, I'd make sure you are spec'd correctly for the number of instances you are monitoring.  For more specific information or analysis, I'd recommend opening an ticket with their tech support folks from your customer portal as this could get involved.  8 )