How to Configure a Clustered Storage Space in Windows Server 2012

How to Configure a Clustered Storage Space in Windows Server 2012

Rate This
  • Comments 35

This blog outlines the sequence of steps to configure a Clustered Storage Space in Windows Server 2012 using the Failover Cluster Manager or Windows PowerShell®. You can learn more about Storage Spaces here:

http://blogs.msdn.com/b/b8/archive/2012/01/05/virtualizing-storage-for-scale-resiliency-and-efficiency.aspx

Prerequisites

  •          A minimum of three physical drives, with at least 4 gigabytes (GB) capacity each, are required to create a storage pool in a Failover Cluster.
  •          The clustered storage pool MUST be comprised of Serial Attached SCSI (SAS) connected physical disks. Layering any form of storage subsystem, whether an internal RAID card or an external RAID box, regardless of being directly connected or connected via a storage fabric, is not supported.
  •          All physical disks used to create a clustered pool must pass the Failover Cluster validation tests. To run cluster validation tests:
    • Open the Failover Cluster Manager interface (cluadmin.msc) and select the Validate Cluster option.

  •          Clustered storage spaces must use fixed provisioning.
  •          Simple and mirror storage spaces are supported for use in Failover Cluster. Parity Spaces are not supported.
  •          The physical disks used for a clustered pool must be dedicated to the pool. Boot disks should not be added to a clustered pool nor should a physical disk be shared among multiple clustered pools.
  •          Storage spaces formatted with ReFS cannot be added to the Cluster Shared Volume (CSV).

Steps to configure using the Failover Cluster Manager

1.       Add the File Services Role and the File Services Role Administration Tools  to all nodes in the Failover Cluster

2.       Open the Failover Cluster Manager interface (cluadmin.msc)

3.       In the left-hand pane, expand Storage. Right-click on Pools and select New Storage Pool. This will start the New Storage Pool Wizard

4.       Specify a Name for the Storage Pool and choose the Storage Subsystem that is available to the cluster and click Next

5.       Select the Physical Disks (a minimum of three with minimum capacity 4GB each and bus type SAS) for the storage pool and confirm the creation of the pool. The pool will be added to the cluster and brought Online, once created.

6.       The next step is to create a Virtual Disk (storage space) that will be associated with a storage pool. In the Failover Cluster Manager, select the storage pool that will be supporting the Virtual Disk. Right-click and choose New Virtual Disk

7.       This initiates the New Virtual Disk Wizard. Select the server and storage pool for the virtual disk and click Next.  Note that the cluster node hosting the storage pool will be listed.

8.       Provide a name and description for the virtual disk and click Next

9.       Specify the desired Storage Layout (Simple or Mirror; Parity is not supported in a Failover Cluster) and click Next

Note: I/O operations to a CSV mirror space are redirected at the block level through the CSV coordinator node. This may result in different performance characteristics for I/O to the storage, compared to a simple space.

10.   Specify the size of the virtual disk and click Next. After you confirm your selection, the virtual disk is created. The New Volume Wizard is launched if you do not uncheck this option on the confirmation page.

11.   The correct Disk and the Server to provision the disk to should be selected for you. Verify this selection and click Next.

12.   Specify the size of the volume and click Next

13.   Optionally assign a drive letter to the volume and click Next

 

14.   Select the file system settings and click Next and confirm the volume settings. The new volume will be created on the virtual disk and will be added to the Failover Cluster.

Note: The NTFS File System should be selected if the volume is to be added to Cluster Shared Volumes.

 

15.   Your clustered storage space can now be used to host clustered workloads. You can also see the properties of the clustered storage space and the clustered pool that contains it, from the Failover Cluster Manager.

 

Steps to configure using Windows PowerShell®

 

Open a Windows PowerShell® console and run the following steps:

1.       Create a new pool

a.  Select physical disks to add to the pool

$phydisk = Get-PhysicalDisk –CanPool $true | Where BusType -eq "SAS”

b.  Obtain the storage subsystem for the pool

$stsubsys = Get-StorageSubsystem

c.       Create the new storage pool

$pool = New-StoragePool -FriendlyName TestPool -StorageSubsystemFriendlyName $stsubsys.FriendlyName -PhysicalDisks $phydisk -ProvisioningTypeDefault Fixed

d.      Optionally add an additional disk as a HotSpare
$hotSpareDisk =
Get-PhysicalDisk –CanPool $true |Out-GridView -PassThru

Add-PhysicalDisk -StoragePoolFriendlyName TestPool -PhysicalDisks $hotSpareDisk -Usage HotSpare

 

2.        Now create a Storage Space in the pool created in the previous step

a.  $newSpace = New-VirtualDisk –StoragePoolFriendlyName TestPool –FriendlyName space1 -Size (1GB)  -ResiliencySettingName Mirror   

 

3.       Initialize, partition and format the Storage Space created

a.  $spaceDisk = $newSpace | Get-Disk

b.  Initialize-Disk -Number $spaceDisk.Number -PartitionStyle GPT

c.  $partition = New-Partition -DiskNumber $spaceDisk.Number -DriveLetter $driveletter -size $spaceDisk.LargestFreeExtent

d.  Format-Volume -Partition $partition -FileSystem NTFS

 

4.       Add the Storage Space created to the Cluster

a.  $space = Get-VirtualDisk -FriendlyName space1              

b.   Add-ClusterDisk $space

 

Note:

  • Clustered Spaces can also be created using the Server Manager:

  • You can find a full end to end Windows PowerShell® sample on setting up a file server cluster with Storage Spaces here.

Troubleshooting tips:

If you come across any of the following errors while attempting to add a storage pool to the cluster please review the Prerequisites section at the beginning of this blog to determine which requirement was not met:

Failed to add storage pool to cluster – {XXXXXXX-XXXX-XXXX-XXXX-XXXXXXX}

No storage pool suitable for cluster was found. 

 

 

Thanks!

Subhasish Bhattacharya                                                                                                               
Program Manager                                                                                                          
Clustering & High Availability                                                                                      
Microsoft

Leave a Comment
  • Please add 5 and 4 and type the answer here:
  • Post
  • What sense does it make to cluster local disks?

  • This is not with local disks.  It is a new scenario where you use simple low cost external shared SAS JBOD storage with Spaces to create resilient storage that is clustered and commonly accessible to all nodes.

    Thanks!

    Elden

  • From the article:

    "Note: I/O operations to a CSV mirror space are redirected at the block level through the CSV coordinator node. This may result in different performance characteristics for I/O to the storage, compared to a simple space."

    What does that mean exactly?

  • I/O operations to a CSV mirror space are sent over SMB 3.0 to the CSV coordinator node (the node on which the NTFS volume is mounted) and is then written to the NTFS volume. CSV is said to operate in redirected mode i.e. I/O is not directly written to the NTFS volume on the local node. In Windows Server 2012, we have introduced a new mode of redirection, at the block level, where I/O passes through the CSVFS proxy file system stack on the node on which it is initiated but is written directly to NTFS volume on the CSV coordinator node. I/O thus avoids passing through the file system stack twice.

  • Are parity spaces also in redirected mode? Are there any whitepapers on the real performance cost this has?

  • Parity layouts are not supported on clustered storage pools. There is a hard block against this configuration.

    At the moment, there are no whitepapers published analyzing the I/O performance of a mirrored space on CSV.

  • We usually use FC or iSCSI to connect to the external storage. If storage pools are currently supporting only SAS(which is internal), does that mean that the storage pool feature cant be efficiently used for failover clusters, say for Disaster Recovery at this point ?

  • Yes, in Windows Server 2012 you are able to leverage Clustered Spaces only for SAS connected JBODs.

    To explain the philosophy/rationale behind this: In Windows 8 (equally applicable to Windows Server 2012), Storage Spaces enables delivery of a new category of highly capable storage solutions at dramatically lower price-points & maximized operational simplicity. In doing so, we have strived to ensure a predictably consistent usage experience (irrespective of storage component hardware failures) while delivering excellent performance. Our internal testing exposed complex interactions when combining Spaces with “storage arrays” connected via iSCSI and/or FC. Therefore, for Windows 8 (Windows Server 2012), we are constraining Spaces based deployments to SAS connected JBODs (for business critical deployments), and SATA/USB connected disks (for home/enthusiast/small-business customers).

  • Hello,

    Describes hardware information? Thank you.

    If you must use an external disk, hard drive needs to be SAS dual-port it? Thank you.

    Build my own NAS

    (Own)

    And cluster reliability and low cost.

  • I don't see it explicitly stated so I have to ask:  because we are working with a storage pool in a cluster ... does that mean the direct attached SAS can be attached to any node in the cluster?  Or are we making a storage pool per node?  So can I make a 3 node cluster with an extra SAS drive per node and then still create and present a single storage pool?  

    I'm assuming no ... but you never state each drive needs to be on a single node.

  • How about going over the hardware requirements for this, what kind of JBOD is required and how should it be attached to the cluster servers?

  • Hi Justin, You are correct in that the there is a requirement that necessitates shared storage i.e. the minimum of 3 physical required drives to create a storage pool needs to be visible from a minimum of 2 cluster nodes in Windows Server 2012.

  • Hi Ben,

    The *hardware* requirement is that your configuration needs to pass the Cluster Storage validation tests. We have tested and worked with the major vendors to ensure that they are compatible however the only definitive test is passing Cluster validation.

  • Can you explain this "Layering any form of storage subsystem, whether an internal RAID card or an external RAID box, regardless of being directly connected or connected via a storage fabric, is not supported?"

    What does that mean exactly? I have 2 systems that have a SATA controller integrated on the system board. I'm using those drives for the OS. I want to connect the required SAS drives using a SAS controller I can install into an available PCI or PCIe slot. Will that work? If not, can I use the SAS card at all and use SAS drives only?

  • i also am curious on the hardware side of things. you mention that you are using a DAS, but never shown how to make sure that both nodes see the exact same storage. I know that probably is a different request but not being familiar with clustering DAS (have worked with iscsi clusters), is the a connecting fail-over cluster to DAS walk-through you could recommend? the instructions you provide assume the reader has already setup the fail over cluster connection to the DAS and just needs to click on create Pool.

Page 1 of 3 (35 items) 123