No announcement yet.

DFS Without Replication?

  • Filter
  • Time
  • Show
Clear All
new posts

  • DFS Without Replication?

    I have a question. About two years ago we went through a file server consolidation project. During this project we used the File Server Migration Tool to move the data and recreate the shares. This worked well. Our idea in this project was to go from all these small file servers to a couple big ones. The problem is that one of the big ones is simply too big. Right now it is about 6 TB of consumed space and we have a hard time backing up the server and restore times are going to be way too long. Obviously this isn't acceptable.

    So simple answer is the split it up. Of course the problem is that we will have to update the shares. There is no way around this. We did this two years ago and have to do it again now. However I suggested that we use DFS then and I am pushing harder for it this time. While I agree with the argument that we don't need DFS for the fault tolerance. Now it would be nice, but for budget reasons we can't afford another 6 TB of disk on our SAN for permanent replication. I can get the space for a migration period but in the long run I have to give it back.

    My concern is that this data is mainly department shared folders. They are always growing. Today it isn't a big deal as we just add another volume to the server and move a department over to the new, larger volume. This doesn't change share names so it doesn't break anything. My issue is once we split up the servers, if we want to keep them at about 1TB in size we are going to have to move departments around in the future. Meaning that we have to break things and change GPOs and rewrite applications that use these shares all over again. If we were to use DFS I would image we could then leverage DFS to migrate the data to the new location, once everything is in sync, kill the old location and nothing would change. Sure we would be replicating a department for a while but that shouldn't be a big deal. I wouldn't be talking about 6TB of data but more like 400GB.

    So what am I missing? Is there any reason we shouldn't use DFS in a manner like this? From my understanding of what I have read there should be no issues. All the servers involved with this project are either Server 2008 (R1) or Server 2008R2. I don't see why this won't work but I am getting a lot of push back from management right now about DFS as they have it stuck in their heads that we have to dedicate enough space to replicate all the data.


  • #2
    Re: DFS Without Replication?

    I set up DFS on our network for convenience as you describe.

    The 'shares' are virtual so, as you say, when you need to move data to a new machine you simply update the location on the DFS server - takes about a minute and no-one is any the wiser. Saves you and your staff a lot of time reconfiguring their machines.

    We do not use replication because we simply do not have the extra hardware to be able to do that. Replication is configured independently of the 'shares' (the namespaces), so by ignoring it, data is not replicated between servers. This does not affect the ability to use DFS.

    However, you need to be aware that if you use a server to host DFS, and your data is spread acros two or more different servers, if the DFS server goes down, so do your shares. Even if the machines are up and accessible, users will not be able to connect via DFS. They can, of course revert to mapping a drive or using UNC paths to access those machines.

    But - if you use the DFSUtil tool to backup the namespaces it is a cinch to restore that data to another server. Again, this will be quicker than reconfiguring client computers.
    A recent poll suggests that 6 out of 7 dwarfs are not happy