Creating a backup network - Veritas Cluster Server

This is a discussion on Creating a backup network - Veritas Cluster Server ; We have 2 x 2-node clusters (one for F&P and one for Exchange). All nodes have teamed NIC's connected to a single switch. We have an LTO Autoloader attached to the F&P cluster and BackupExec 8.6 cluster-aware version installed a ...

+ Reply to Thread
Results 1 to 2 of 2

Thread: Creating a backup network

  1. Creating a backup network

    We have 2 x 2-node clusters (one for F&P and one for Exchange). All nodes
    have teamed NIC's connected to a single switch.

    We have an LTO Autoloader attached to the F&P cluster and BackupExec 8.6
    cluster-aware version installed a running smoothly backing up the two
    clusters and two stand alone machines.

    We have recently put a Gig NIC in each of the four nodes and these are
    connected to a separate Gbix switch. The two standalones do not have Gig
    cards.

    Question is this:
    How can we force the backup to occur over the Gig cards, or force it to
    occur over the 10/100 cards for that matter?
    A consideration is that we may run backup jobs to backup folders on the
    Cluster and backup that data during working hours.

    Now feel free to slate the design here, I didn't created it (inherited along
    with numerous other head-scratching and nonsensical designs). I just wonder
    if someone has thought of a real clever thing to do that I just don't get.

    As far as I can see it the only way to use the Gig cards effectively is to
    make them the cards for the cluster resource (which means rebuilding the
    cluster?) OR using them as a separate cluster resource with separate IP addr
    (on the same subnet) to run the backup during the night.

    To be honest any and all feedback would be most appreciated.

    Cheers
    Dave



  2. Re: Creating a backup network

    Dave,

    If you post this to the Backup Exec Windows Group and the VCS WindowsNT
    groups (veritas.vcs.windowsnt.english; veritas.backupexec.windowsnt.english)
    you will probably get more feedback, this newsgroup applies mostly to the
    Unix versions of VCS. Technical support would be able to shed more light on
    this as well.

    Backup Exec 8.6 does have a new feature that allows targeting which
    interface to backup across, but I believe that is only from the the Backup
    Server side not the client.

    You didn't say which version of VCS you are running but if you are running
    2.0 on Windows 2000 configuring the new network cards inside VCS shouldn't
    present a difficult problem.

    For backing up across a different NIC on the same subnet, you might want to
    rethink that strategy after reviewing Microsoft Technet. There are some
    definite problems that could occur if you don't configure everything
    correctly and also a strong likelihood that you won't get the desired
    results.

    Better solution might be to create a separate VLAN in another subnet to
    force backups to use another path so that the production network is not
    affected by the backup during the day. When configuring the other NIC on
    the other subnet be sure that no routing is configured between the subnets.
    VCS could be used to online the other NIC and IP resource and could reside
    in the same group if necessary, just don't make the resources critical on
    this NIC or IP resource. The new LanMan in VCS 2.0 can broadcast the
    Virtual name out on the backup network as well so that Backup Exec can see
    the NetBios Name on your new subnet.

    Hope this helps.

    -kk

    "Dave Wilson" wrote in message
    news:3d22c2f0$1@hronntp01....
    > We have 2 x 2-node clusters (one for F&P and one for Exchange). All nodes
    > have teamed NIC's connected to a single switch.
    >
    > We have an LTO Autoloader attached to the F&P cluster and BackupExec 8.6
    > cluster-aware version installed a running smoothly backing up the two
    > clusters and two stand alone machines.
    >
    > We have recently put a Gig NIC in each of the four nodes and these are
    > connected to a separate Gbix switch. The two standalones do not have Gig
    > cards.
    >
    > Question is this:
    > How can we force the backup to occur over the Gig cards, or force it to
    > occur over the 10/100 cards for that matter?
    > A consideration is that we may run backup jobs to backup folders on the
    > Cluster and backup that data during working hours.
    >
    > Now feel free to slate the design here, I didn't created it (inherited

    along
    > with numerous other head-scratching and nonsensical designs). I just

    wonder
    > if someone has thought of a real clever thing to do that I just don't get.
    >
    > As far as I can see it the only way to use the Gig cards effectively is to
    > make them the cards for the cluster resource (which means rebuilding the
    > cluster?) OR using them as a separate cluster resource with separate IP

    addr
    > (on the same subnet) to run the backup during the night.
    >
    > To be honest any and all feedback would be most appreciated.
    >
    > Cheers
    > Dave
    >
    >




+ Reply to Thread