We have 2 identical servers (DR & Viewonly), both restore the same db files copied down from production daily. Over the past 2 years, one server (DR) has consistenly restored the database in 1/4 of the time as the other (ViewOnly). The ONLY difference is the name of the database in DR is the same as production whereas VIewOnly restores to a different db name.

Is it possible the different database name is slowing down the restore process?

Logs:
DR
3/13/2011 7:55:25 PM: RESTORE DATABASE successfully processed 51800589 pages in 2118.718 seconds (200.286 MB/sec).

DR - almost a year ago
5/9/2010 3:59:47 AM: RESTORE DATABASE successfully processed 38458043 pages in 1479.035 seconds (213.009 MB/sec).

View only - today
3/15/2011 8:56:28 AM: RESTORE DATABASE successfully processed 49144879 pages in 13278.791 seconds (30.318 MB/sec).

View only - 1 year ago
3/15/2010 RESTORE DATABASE successfully processed 64603668 pages in 10756.321 seconds (49.202 MB/sec).

View only - 2 years ago
3/16/2009 RESTORE DATABASE successfully processed 39596151 pages in 6596.895 seconds (49.170 MB/sec).
davidbanner2k
0

Comments

5 comments

  • petey
    The database name shouldn't be a factor in the restore process speed.

    If you ran RESTORE VERIFYONLY on the same file on both servers, is the duration also different?

    Thanks.
    petey
    0
  • davidbanner2k
    Thanks Peter, I will give this a try and let you know results.
    davidbanner2k
    0
  • davidbanner2k
    Rebuild the database with same name and tried verifyonly method. Still no luck.
    Any other ideas?
    davidbanner2k
    0
  • petey
    Are you saying that verifying the backups on the read-only server is still significantly slower than on the disaster recovery server? Are the 2 servers identical in their hardware configuration, especially their CPU and disk configurations?
    petey
    0
  • davidbanner2k
    yes.
    The servers are identical in terms of memory/cpu/OS. They are both using SAN, the only difference being letters/labels because the storage team didn't configure identical. But the mdf/ldf/ndf files are allocated across in the same manner. I do have an open ticket with the SAN team to do a health check, pending analysis right now.
    Thanks again.
    davidbanner2k
    0

Add comment

Please sign in to leave a comment.