Post reply

Backup job runtimes vary greatly

  • July 8, 2009 at 10:10 am

    #134163

    Please help if you can.

    My environment: Clustered SQL Server 2000 SP4/W2K Server 2003 SP1. Each cluster member has 2 fiber mapped drives to a SAN, one for data one for logs. My full DB backups are going to a UNC path \\MYFILESERVER\DBBACKUP. This UNC path is NAS.

    My backup file size is about 25 GBs. My backup times are varying between 55 minutes and 5 hours+ ! This backup is run the same time every day and user activity is consistent during this time.

    I think my issue may have to do with going from SAN through network to NAS.

    Can anyone provide any advice on how to determine what is causing these discrepancies? Are there any alternate architectures that may help my backup performance in this scenario? Would it be a bad idea to backup to the mapped fiber drives given the redundancy of the SAN?

    Thanks, Emily

  • July 9, 2009 at 9:28 am

    #1022482

    Emily,

    When you say user activity is consistent, are you referring to the SQL Server or the SAN? I have seen similar issues with varied backup times related directly to SAN performance/load before. One thing I did was run the backup to local storage, then copy the file using a batch job to the SAN.

    Dave

    Dave

  • July 9, 2009 at 9:30 am

    #1022483

    It's never a good idea to back up to mapped drives across a network. A backup is NOT like a file copy. It does not retry, and has low tolerances for network issues. My guess is the SAN might smooth out the bumps and allow the backup to complete by using caching.

    You are much better off backing up to a local OS drive, on the SAN or not, and then running a file copy.

    The same notes for backups apply. Use a different drive other than the one for data.

  • July 13, 2009 at 1:48 pm

    #1024129

    >>You are much better off backing up to a local OS drive, on the SAN or not, and then running a file copy.

    This is the direction I would like to go. We are going to add some additional storage, recognized by the cluster, on the SAN and then backup to the SAN storage.

    When you say run a file copy how is this done? Is this a command line copy that would be run as the last step in a backup job or is there sperate software that runs this file copy job.

    Thanks

  • July 15, 2009 at 11:22 pm

    #1025481

    File copy using VBScript or Powershell to move the files. VBScript with the FileSystemObject is pretty easy to write to move the files, and there should be scripts here on the site to help with that.

    You can put this in a batch file and call it from SQL Agent, or in a DTS package and use that.

  • July 16, 2009 at 7:38 am

    #1025637

    I have a step in my backup jobs in SQL Server Agent that uses the CMDEXEC job type and does a DOS COPY to copy the file to a network share.

    Jack Corbett
    Consultant - Straight Path Solutions
    Check out these links on how to get faster and more accurate answers:
    Forum Etiquette: How to post data/code on a forum to get the best help
    Need an Answer? Actually, No ... You Need a Question

  • July 21, 2009 at 12:23 pm

    #1027839

    Thanks all, I would say this is resolved, backup times have been greatly reduced by backing up to a different array of disks on the SAN.

Viewing 7 posts - 1 through 6 (of 6 total)

You must be logged in to reply to this topic. Login to reply