none
DPM 2010 Bare Metal Restore Backup RRS feed

  • Question

  • Specifically, the Bare Metal Recovery backups are KILLING us.  I’ve tried setting DPM to compress the BMR backups, but that doesn’t seem to be helping.  Our plants have T1 connections to us, but we just can’t seem to get enough throughput to get the backups done in a timely fashion (or at all!). Anything I should be looking at or tweaking to get better performace?

     

    Thanks.

    Monday, October 31, 2011 8:20 PM

Answers

  • Hello,

    I am curious as to what all is included for your server for BMR critical volumes.

    To see which volumes will be included in a BMR (which includes systemstate) backup, you can open an administrative command prompt and run the following command:

    C:\>wbadmin start backup –allcritical –backuptarget:<some valid target path in the form of \\servername\sharename>

    Before the confirmation to start the backup, wbadmin will list all the critical volumes in the system that will be included of the BMR backup. This will help you plan the DPM replica volume size ahead of time. Again you can select "N" for no it'd be interesting to see though what is showing up in the list for critical volumes.


    Shane

    Friday, November 11, 2011 1:50 PM

All replies

  • Hello,

    I'd stagger your protection groups to not synch at the same time, have the BMR synchs take place after hours and use compression on the wire (your already doing this).  These are all good places to start.


    Other performance articles:

    DPM Managing Performance:  http://technet.microsoft.com/en-us/library/ff399471.aspx
    How DPM Operations Affect Performance:  http://technet.microsoft.com/en-us/library/ff399315.aspx
    DPM and Memory:  http://technet.microsoft.com/en-us/library/ff399244.aspx
    Performance Counters:  http://technet.microsoft.com/en-us/library/ff399366.aspx
    Improving Performance:  http://technet.microsoft.com/en-us/library/ff399571.aspx
    Managing DPM Performance on a WAN:  http://technet.microsoft.com/en-us/library/ff634226.aspx
    How Protection Groups Changes Affect Jobs:  http://technet.microsoft.com/en-us/library/ff399316.aspx


    What other datasources are included in these PGs?

     

    Thanks
    Shane

     

    Tuesday, November 1, 2011 1:11 PM
  • Thanks for the reply.  I’ve looked over the various links and articles.  I’ve modified the page file as per recommendation here, but don’t believe that relates to the trouble we’ve been having.  I’ve watched the performance counters for a while and all are well within the specifications mentioned.  Most of the jobs are already staggered, but when we’re talking about a Bare Metal Recovery backup that takes days to complete (if it ever does), staggering the jobs would make no difference.  Lastly, I looked at the article about performance over a WAN and am not certain it relates.  The article is VERY vague and the only example provided is for a 100 mbps connection!  If I had that kind of throughput on my WAN, I don’t think I’d be worrying too much about bandwidth and backup jobs not completing.  Without some real-world examples and explanations as to what the various settings should be and what they are meant to accomplish, I’m not able to attempt them.

     

    I don’t have a problem with waiting days or even weeks to get an initial BMR.  The real problem is that, once we get it, the recovery points appear to generate backups just as big as the initial job.  As a result, we can never seem to get caught up.  This seems strange to me in that my understanding was that DPM would only attempt to replicate changed blocks and not the entire backup every time.  Take, for example, our Hyper-V backups at the plants.  The initial backup is painful and takes days, but once we have it, DPM is able to keep up with changes without difficulty.  How is it that a Hyper-V backup only sends changed blocks, but a BMR backup of the same host seems to want to re-send the entire backup each time?

     

    I must have mis configured something....

    Tuesday, November 1, 2011 8:28 PM
  • Hello,

    I am curious as to what all is included for your server for BMR critical volumes.

    To see which volumes will be included in a BMR (which includes systemstate) backup, you can open an administrative command prompt and run the following command:

    C:\>wbadmin start backup –allcritical –backuptarget:<some valid target path in the form of \\servername\sharename>

    Before the confirmation to start the backup, wbadmin will list all the critical volumes in the system that will be included of the BMR backup. This will help you plan the DPM replica volume size ahead of time. Again you can select "N" for no it'd be interesting to see though what is showing up in the list for critical volumes.


    Shane

    Friday, November 11, 2011 1:50 PM