Forum Discussion

Michael_Fisher's avatar
11 years ago

Recommended client policy per duration

Running Netbackup 7.6.0.2. Will I get better performance by grouping clients with a long backup window together in the same policy or mix a long backup window client with short backup duration clients?

  • When I size backup jobs I do it by throughput to the target media server.  This is usually dicatated by network, disk speed, and client change data.  More often network speed.  

    First off use accelorator if you can.  Not having to send the files to the media server helps on a thin network where data change rates are small.  I have media servers that run 40 jobs at the same time and some that run only 10.  

    I classify a job as a stream to a client.  Streams = kb pers second.  I try to keep my nics media server nics near full speed during backup windows without slowing down the source client.  This ensures that I get data as fast as clients can give it and I get the most clients done as I possibly can.  

    I love the week at a glance report from ops center for this info.  Each backup will show you how much "new" data the job got.  Just hold the mouse over the success icon (from the ops center interface).  

     

    Sully

     

     

7 Replies

  • When I size backup jobs I do it by throughput to the target media server.  This is usually dicatated by network, disk speed, and client change data.  More often network speed.  

    First off use accelorator if you can.  Not having to send the files to the media server helps on a thin network where data change rates are small.  I have media servers that run 40 jobs at the same time and some that run only 10.  

    I classify a job as a stream to a client.  Streams = kb pers second.  I try to keep my nics media server nics near full speed during backup windows without slowing down the source client.  This ensures that I get data as fast as clients can give it and I get the most clients done as I possibly can.  

    I love the week at a glance report from ops center for this info.  Each backup will show you how much "new" data the job got.  Just hold the mouse over the success icon (from the ops center interface).  

     

    Sully

     

     

  • There is not much difference in running 1 or 100 jobs to backup 100 clients for instance. As long as your settings are set to allow as many jobs as the media server can handle go active at the same time. As posted above, look at the media server's capability.

     

    From a management point of view you would want as few policies as the environment's requirements allow.

  • Thanks for the responses. I don't see a report named "week at a glance" in ops center. Where can I find or create it?

  • in ops center, 

     

    Reports >> Stats & Success Rate >> status

     

    I'm on ops center for 7.5.5.  If the report is not there it may be because I bought analytics.. 

     

    Sully

  • When doing vmdk backups you do not want many clients being backed up at the same time. You want the individual client backup to complete as quickly as possible not the group of clients to complete quickly. This requirement  is due to needing to shorten the time of quiescing the client during the backup.