1 Reply Latest reply on Jun 8, 2014 2:15 PM by Bill Robinson

    BSA Snapshot Efficiency

      In term of running Snapshots, what piece of the environment takes on the blunt of the workload, besides the database server.  Is it the App Servers, the Agent?


      Obviously the Agent provides a communication channel for the App Servers and the Target, and the App Servers have all of the rules.  So with that said, are the App Serves doing the scans themselves or do the App Servers send down “directives/Instructions/content” down to the Agents and the Agents run the snapshots?


      Reason I am asking, is, I am trying to improve on the efficiency on how our Snapshots are ran.  I know that most of this work will ultimately come from tweaking the templates (what it is we are scanning), but aside from that, from a resource level, what areas in our environment can I scale out to help improve the process.


      For example, I have a template, for good, bad, worse, using this single template it takes 30 hours to run across our entire environment by running 150 parallel targets using 3 Job Servers.

      1. Would simply adding additional Application Servers and increasing the number of targets I run in parallel give me that much better results, or is the bottleneck always going to be at the Database Level?
      2. If the Server is the one doing the scans on the targets, will network resources be a factor?  Are snapshots going to be slower when ran on Servers in other locations/sites (vs in the same datacenter as the App Servers).
      3. If I need to add resources to our database server, I assume it would need to be CPU?
        • 1. Re: BSA Snapshot Efficiency
          Bill Robinson

          Appservers do the delta calculation.  There was this ‘edge processing’ setting back in 6.x that had the target process the delta but that’s long gone…


          Depending on what you are gathering (eg, file contents), gathering across a wan will of course be slower than gathering locally.  Not much you can do about that.


          I’d start looking at db performance reports (eg, awr, addm, sql trace), and appserver os reports and see if you are running into performance issues w/ either of those before adding more job servers.