Quorum device in a stretch cluster - Veritas Cluster Server

This is a discussion on Quorum device in a stretch cluster - Veritas Cluster Server ; The following text is applicable to a cluster 2 node Sun Cluster with 2 disk arrays, 1 node/array per site. "Special considerations also need to be given to placement of voting disks. For example, in case of network connectivity problems ...

+ Reply to Thread
Results 1 to 2 of 2

Thread: Quorum device in a stretch cluster

  1. Quorum device in a stretch cluster


    The following text is applicable to a cluster 2 node Sun Cluster with 2 disk
    arrays, 1 node/array per site.

    "Special considerations also need to be given to placement of voting disks.
    For example, in case of network connectivity problems between the 2 sites,
    building a proper quorm mechanism to avoid split-brain situations is important
    consideration in this configuation. Firstly the mirroring solution needs
    to be cluster-aware. Secondly, if the quorum device is located in one of
    the 2 sites, and that site is struck with a disaster, the entire cluster
    is brought down and must be manually reinstated. For maximum availability,
    the quorum device ought to be placed at a 3rd location, which increases the
    operational complexity as well as the cost."


    My question is, is this design problem also affecting VCS? Or does VCS have
    a clever workaround to avoid requiring a 3rd site?

  2. Re: Quorum device in a stretch cluster

    Don't know SUN cluster at all.


    The only time you need disks with VCS, is when you have Oracle RAC. With
    RAC, you will need coordinator disks (used in voting in a membership
    change).


    In a normal VCS situation, no quorum disks needed




    Steve wrote:
    > The following text is applicable to a cluster 2 node Sun Cluster with 2 disk
    > arrays, 1 node/array per site.
    >
    > "Special considerations also need to be given to placement of voting disks.
    > For example, in case of network connectivity problems between the 2 sites,
    > building a proper quorm mechanism to avoid split-brain situations is important
    > consideration in this configuation. Firstly the mirroring solution needs
    > to be cluster-aware. Secondly, if the quorum device is located in one of
    > the 2 sites, and that site is struck with a disaster, the entire cluster
    > is brought down and must be manually reinstated. For maximum availability,
    > the quorum device ought to be placed at a 3rd location, which increases the
    > operational complexity as well as the cost."
    >
    >
    > My question is, is this design problem also affecting VCS? Or does VCS have
    > a clever workaround to avoid requiring a 3rd site?


+ Reply to Thread