DFS in Windows 2003 environment Scenario

Discussion in 'File Systems' started by a_user, May 29, 2005.

  1. a_user

    a_user Guest

    Hello gang,

    Please excuse my ignorance on the subject of my question, however I am a
    little unclear on the the best away to approach my goal.

    I would like to implement redundancy of file shares within my environment.
    I am planning two servers; fs01 and fs02 as we move into a newly designed
    2003 AD environment. I was contemplating using DFS to ensure data
    availability by eliminating any one point of a failure if either of the two
    file servers are offline. However my data will be stored on a SAN. What I
    am unclear about is the necessity of the using FRS. Everything I read
    indicates that file replication services is used to replicate the shared data
    between each of the servers in the DFS directory that are configured as
    target hosts. I cant understand why this would be necessary in my situation
    though and infact I do not want this to occur.

    I simply want the two file servers to provide a frontend to the data hosted
    on the SAN, but in such a way that users are not mapping specifically to one
    single server name. Therefore if either of the two servers is offline the
    client will automatically use the second server's link to the data to
    continue accessing the data stored on the SAN.

    I would like to avoid the use of clustering in this case for such a simple
    solution given the central repository of data in the first place.

    Will implementing DFS allow me to implement this solution the way I propse
    or is there a catch in there somewhere that will prevent me from seeing the
    results I am trying to achieve.

    Thank you.
    a_user, May 29, 2005
    1. Advertisements

  2. The idea behind using DFS is to have users map to a single link and get
    redirected to one of multiple link targets that contain the actual data.
    These link targets must contain identical data to make sure the user gets a
    consistent view of the data in the event that one of the links fails.

    DFS will not work if you are not willing to duplicate the data and keep it
    up to date across all link targets.
    Here is the issue.
    You cannot have two file servers accessing the same shared LUN at the same
    time and expect there not to be volume corruption. Windows OS demands
    exclusive access to the LUNS presented to it.

    Clustering is the solution here. It will provide the fault tolerance you
    are looking for without having to incur the cost of duplicating the data.
    The clustering code will ensure only one node will have exclusive access to
    a LUN at any given time.
    Glenn LeCheminant, Jun 1, 2005
    1. Advertisements

  3. a_user

    a_user Guest

    Hello Glenn,

    Thank you for taking the time reply. That does make sense then. I was
    hoping to avoid the necessity of having to duplicate data in that way,
    however the point you raise about accessing the same data at the same time
    causing problems makes sense.

    Now I know and understand why frs is required to achieve the results and not
    as simple as I had hoped for.

    Thanks again.
    a_user, Jun 2, 2005
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.