Creating a Cluster Shared Volume RRS feed

  • Question

  • Hi all,
    So I am new to clustering and I have 6 identical servers. I am trying to create cluster to run my vms on.
    I am following the steps in the following link:
    My question is for my 6 servers i need to dedicate one of them as an ISCSI target server and the other 5 servers (nodes) can use iscsi initiator to use the CSV on the target server. But what if i want to add physical harddrives to my CSV, how would i do that?
    Or I can only create volumes from physical drives connected to the ISCSI target server.

    Also I feel bad losing one whole server for the clustered shared volume as an ISCSI target server is there anyway I can do this in a better way, noting that I have a file server on my network.

    Let me know if I am missed any information. 
    Wednesday, May 30, 2018 9:43 PM

All replies

  • Hi,

    There is a article for you,please refer to it.

    Use Cluster Shared Volumes in a Failover Cluster


    Best Regards,


    Please remember to mark the replies as an answers if they help.
    If you have feedback for TechNet Subscriber Support, contact tnmff@microsoft.com

    Thursday, May 31, 2018 5:40 AM
  • "for my 6 servers i need to dedicate one of them as an ISCSI target server and the other 5 servers (nodes) can use iscsi initiator to use the CSV on the target server."

    Using just a single server as the iSCSI target makes that a single point of failure for your VM cluster.  It would be better to use two of the servers as a clustered iSCSI target to ensure availability of the storage.  Actually, I would use two servers to create a cluster to serve SMB shares.  Setting up and using SMB shares is easier than setting up an iSCSI environment and the performance is at least as good, and possibly better.


    "Or I can only create volumes from physical drives connected to the ISCSI target server."

    If you are going to use iSCSI, yes, you will access the storage on the iSCSI target and create your CSVs using that storage.


    Thursday, May 31, 2018 11:25 AM
  • Firstly, thanks for your reply.
    I think i did not make my question clear, to clarify I do not want to a dedicate single server as iscsi target and not being able to have 6 nodes within the cluster.
    So is dedicating atleast one of the 6 servers for being an iscsi server the best way to do it?
    I also have harddrive an all 6 servers so i want to add the hard drives on the other 5 servers to this clustered shared volume, is that doable?
    I am also running windows server 2016 on all of them if this is useful info.

    Thursday, May 31, 2018 2:12 PM
  • You have two options to use the local storage on all the servers as storage for use by the cluster.  In Windows Server 2016, Microsoft introduced Storage Spaces Direct (S2D).  This allows similarly configured hosts to be joined to a cluster and all the local storage to be shared to the cluster.  In order to do this, the first thing you should read is the hardware requirements - https://docs.microsoft.com/en-us/windows-server/storage/storage-spaces/storage-spaces-direct-hardware-requirements.  Note that while it is possible to 'roll your own' configuration, you absolutely must ensure that the individual components are supported by the hardware vendor for use in a S2D configuration.  The easiest way to ensure this is to purchase a system from a vendor who has already configured a system and performed a significant amount of testing to certify that the system and all its components work (https://docs.microsoft.com/en-us/windows-hardware/test/hlk/testref/partner-guide-to-wssd-certification).  If you opt to roll your own, that means that you will be performing that sort of testing on your own.  You can find a list of those vendors who have certified systems here - https://www.microsoft.com/en-us/cloud-platform/software-defined-datacenter.

    Another way to use local storage across the cluster is to use a third party solution, such as Starwind.  Then you split your support between Microsoft and Starwind.

    FYI, your initial assumption of the ability to place all nodes in a cluster and to use one or more nodes as storage nodes is not a valid configuration.  Other than the new S2D configuration, Microsoft Failover Clustering does not support shared storage coming from the nodes of the cluster accessing the storage.  Shared cluster storage must come from a different source, such as a SAN, SAS array, iSCSI target, or SMB 3.0 (or later) shares.  In each of these cases, it is recommended that the source of the shared storage be highly available so that it does not present a single point of failure.


    Friday, June 1, 2018 12:14 PM