none
Clustering existing servers running dfs

    Вопрос

  • I'm looking at creating a first cluster and need some advice. Got 1 currently production server and one non production brought on line for this. Both are running Server 2008 R2 Enterprise.

    The production server though is running DFS which we are using for file shares, roaming profiles, folder redirection etc. How will that need to be changed or be affected by creatig a cluster. I was hoping to get that done, then have this cluster running clustered DFS and also clustered DHCP. Not worried about DHCP though as it's up and running somewhere else.


     
    • Изменено Carito 23 марта 2012 г. 20:25
    23 марта 2012 г. 20:16

Ответы

  • Hi,
    I am not a storage professional. However, as far as I know, you can simply add a new computer to an existing DFS, configure replication for it. Then if one server target fails, the DFS will continue to work without interrupt. So, in a way, a DFS is a kind of cluster.
    By the way, if you want to build a DHCP Failover Cluster, you can check the following guide.
    DHCP Step-by-Step Guide: Demonstrate DHCP Failover – Clustering in a Test Lab
    http://technet.microsoft.com/en-us/library/ee405263%28v=ws.10%29.aspx

    Vincent Hu

    TechNet Community Support

    • Помечено в качестве ответа Vincent HuModerator 9 апреля 2012 г. 15:39
    26 марта 2012 г. 6:57
    Модератор
  • Hiya,

    Ok, basically what you want is to move your current DFS to a clustered file service. (Proberly to avoid having the file storage consuming 2x disk space.)
    First of all, you dont cluster DFS. The DFS is one way of providing high availability(and other functionality) to your fileservices. You choose either to cluster your service or create DFS. There are ofcourse pros and cons with both, you should consider your requirements and options.

    If you choose clustering, the oversimpoified approach is:
    Create your first node in the file service cluster, migrate data. And add the second node to the cluster could be an approach. The cluster will have a different name than the DFS. This will affect any references on the current (DFS)name.

    This checklist might aid you.
    Checklist: Create a Clustered File Server
    http://technet.microsoft.com/en-us/library/cc753969.aspx

    Vincent allready provided reference to setup DHCP failover server.

    Failover Cluster Step-by-Step Guide: Configuring a Two-Node Print Server Failover Cluster
    http://technet.microsoft.com/en-us/library/cc771509(v=ws.10).aspx

    The more specific you can be in your question, the more specific answers we can give :)

    • Помечено в качестве ответа Vincent HuModerator 9 апреля 2012 г. 15:39
    27 марта 2012 г. 7:59

Все ответы

  • Hi,
    I am not a storage professional. However, as far as I know, you can simply add a new computer to an existing DFS, configure replication for it. Then if one server target fails, the DFS will continue to work without interrupt. So, in a way, a DFS is a kind of cluster.
    By the way, if you want to build a DHCP Failover Cluster, you can check the following guide.
    DHCP Step-by-Step Guide: Demonstrate DHCP Failover – Clustering in a Test Lab
    http://technet.microsoft.com/en-us/library/ee405263%28v=ws.10%29.aspx

    Vincent Hu

    TechNet Community Support

    • Помечено в качестве ответа Vincent HuModerator 9 апреля 2012 г. 15:39
    26 марта 2012 г. 6:57
    Модератор
  • Vincent is correct. With DFS you dont need to provide cluster functionality. The wiki article on the topic describes it well.

    Distributed File System (Microsoft)
    http://en.wikipedia.org/wiki/Distributed_File_System_(Microsoft)

    26 марта 2012 г. 7:25
  • Thanks Vincent. I think I didnt ask my question good. I know that when I create a cluster I have to also create a cluster name that and that that is what is used when pointing to the cluser. So I already have DFS running using the \\domain_name\format . If I create a cluster name, which is going to to be different than the DFS name I create, will that affect DFS? Does that make sense?  So I am basically wanting to have a cluster that has DFS, DHCP, and possibly AD Printing.

    So, if I have DFS running on Server 1, and the data is accessed via the server's HBAs on the SAN, and that server goes down how does Server 2 get to the data on the SAN? I know the LUNs are presented to each server already. So wouldnt I need to create a cluster, have a cluster name and have that in DFS somehow not for the DFS links themselves, but to access the data across the HBAs. Does that make sense?

    • Изменено Carito 26 марта 2012 г. 16:19
    26 марта 2012 г. 15:56
  • Hiya,

    Ok, basically what you want is to move your current DFS to a clustered file service. (Proberly to avoid having the file storage consuming 2x disk space.)
    First of all, you dont cluster DFS. The DFS is one way of providing high availability(and other functionality) to your fileservices. You choose either to cluster your service or create DFS. There are ofcourse pros and cons with both, you should consider your requirements and options.

    If you choose clustering, the oversimpoified approach is:
    Create your first node in the file service cluster, migrate data. And add the second node to the cluster could be an approach. The cluster will have a different name than the DFS. This will affect any references on the current (DFS)name.

    This checklist might aid you.
    Checklist: Create a Clustered File Server
    http://technet.microsoft.com/en-us/library/cc753969.aspx

    Vincent allready provided reference to setup DHCP failover server.

    Failover Cluster Step-by-Step Guide: Configuring a Two-Node Print Server Failover Cluster
    http://technet.microsoft.com/en-us/library/cc771509(v=ws.10).aspx

    The more specific you can be in your question, the more specific answers we can give :)

    • Помечено в качестве ответа Vincent HuModerator 9 апреля 2012 г. 15:39
    27 марта 2012 г. 7:59
  • Jesper thanks. I believe your second paragrpah is what I am wanting. I don't have any data to migrate since it all resides on the SAN. In DFS I am only replicating the namespaces not doing any data replication. Kinda two parts here, DFS and Clustering.

    Maybe this helps:

    I have current server #1 actively running DFS and only hosting the namespaces and associated links. All Data is on the SAN accessed over the HBAs.
    In DFS I have a namespace \\domain\namespace_name with targets that point locally to say \\server\U$\share

    I have a second server that I will add DFS to and then add it as a Namespace server to for namespace replication only on the 1st. Again. No data is stored here. This also has the same LUNs presented to it that are presented to server 1.

    So I think what I am getting it is I want to create a cluster so that the data on the SAN is accessible even if server 1 goes down. However all the namespace targets reference server #1 as\\server\u$\share. So what I am getting at is would I create a cluster and use the cluster name in DFS instead of the local path so that if one server goes down the second in the cluster provides availabilty to the SAN across it's HBAs??? Make sense?

    So something like where in DFS I would end up with Namespace \\domain\namespace_name now goes to \\cluster_created_name ?

    27 марта 2012 г. 12:52