Configure Storage Spaces Direct (S2D) on Windows Server 2016

Storage Spaces Direct (S2D) is a new distributed data storage technology that appeared in Windows Server 2016. Due to Storage Spaces Direct, you can turn local drives of several servers into a fault-tolerant, scalable storage protected from failure of both separate disks and entire servers. The cost of this software storage is much less than that of a SAN or NAS due to simplified scaling (up to 16 servers and 400 drives) and the opportunity to use different drives (including SSDs and NVMes) enables significant performance.

What Is Storage Spaces Direct (S2D)

S2D is the further development of Storage Spaces technology and allows to merge the local drives of the Hyper-V cluster nodes into Storage Pools. You can create virtual volumes (disks) on these drives and use them as common Cluster Shared Volume (CSV) to store Hyper-V virtual machines files and SOFS file shares. If you need to expand the storage size, just add a new server or drive to S2D. In general, Storage Spaces Direct is the answer of Microsoft to VMware vSAN.

Storage Spaces Direct Requirements

S2D supports the following storage device types:

• Common HDDs (SAS);
• SATA / SAS SSDs;
• NVMe (Non-Volatile Memory Express) are SSDs connected through a faster PCI Express bus instead of the classic SATA/SAS interface.

Later different types of disks can be combined in various arrays (by speed or size). For example, it is reasonable to locate the cache and application transaction logs on faster NVMe SSDs, and it is better to use slower and less expensive disks to store large files that do not require high performance to access them, etc.

In order S2D to work, you have to create a failover cluster with the following requirements to its nodes.

Requirements to S2D cluster nodes:

1. Windows Server 2016 Datacenter edition;
2. The following components must be installed on the servers: Hyper-V, File Services roles and Failover Clustering feature.
   Note. Don’t forget to disable SMB 1.0: Remove-WindowsFeature –Name FS-SMB1 -Verbose –Restart
3. At least two servers in a cluster (ideally, at least 4 hosts to ensure high fault tolerance);
4. In addition to the system drive, there must be at least one physical disk in each node. All disks that you are going to add to the Storage Spaces Direct must be unformatted (i.e. not partitioned and containing no partition table).

Suppose that you have created a failover cluster of two servers running Windows Server 2016 (you can create it even in a workgroup).

Prior to enabling Storage Spaces Direct, make sure that your disks can be joined into this pool.

Get-PhysicalDisk –CanPool $True | Sort Model

How to Enable Storage Spaces Direct

Activate S2D using the following cmdlet:

Enable-ClusterStorageSpacesDirect

The cmdlet is being processed long enough (about 10 minutes), all available disks and their performance will be analyzed and a cluster pool will automatically be created. Also, two tiers are created automatically: Performance and Capacity, having different failover types: mirror and parity, respectively.

3 types of providing fault tolerance of the data storage are supported:

1. Mirrored (3) – the data are synchronously replicated between 3 (or 2 in minimal configuration) nodes. High reading speed is reached due to distributing the operation between all servers.
2. Parity (2) – the data with parity information are distributed among different disks. Data storage is more efficient since you don’t need to store several copies of the same data.
3. Tiered (1) – a combination of both above mentioned methods.

Note 1. If there appears an error “no disks with supported bus types found to be used for S2D” when running the Enable-ClusterS2D command, it is likely that the bus type (BusType) of your disks is RAID (it is an unsupported configuration from S2D). Let’s check the bus type:
Get-Disk | select Number, FriendlyName, OperationalStatus, Size, PartitionStyle, BusType | sort Number | ft -AutoSize

It is true – in all cases it is RAID. The solution is to update drivers or firmware of the controllers (if you have HP servers, install the latest HPE Support Pack). Check the BusType again. (now it’s changed to SAS).

Also, there is a little trick that allows to change the bus type to SATA for the specific type of controller:

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\arcsas\Parameters
"BusType"=dword:0000000b (instead of 00000008)

Open the Failover Cluster Manager and make sure that Cluster Pool 1 has appeared in the Storage section.

After selecting the pool, you can see what disks it consists of.

If necessary, the pool name may be changed:

Set-StoragePool –FriendlyName “Cluster Pool 1” –NewFriendlyName “S2D”

If you have to create a volume from the specific disks, you can use this command. First, select all LUN 3 disks and collect them into the pool.

$HDDs = Get-PhysicalDisk | ? PhysicalLocation -like "*LUN 3"
New-StoragePool -StorageSubSystemFriendlyName *Cluster* -FriendlyName S2DPool -ProvisioningTypeDefault Fixed -PhysicalDisk $HDDs

Display the list of disks in the pool:

Get-StoragePool -FriendlyName S2D | Get-PhysicalDisk | ft PhysicalLocation

Add a new disk to the pool:

$HDDs = Get-PhysicalDisk | ? PhysicalLocation -like "*LUN 4"
Add-PhysicalDisk -PhysicalDisks $HDDs -StoragePoolFriendlyName S2D

Disks marked as S2D are not displayed in the Disk Management console any more, and it is OK.

If you have different types of drives you can use storage tiering (optional). A mirror-type tier of SSDs is created as follows:

New-StorageTier -StoragePoolFriendlyName S2D -FriendlyName "Mirror_Tier" -MediaType SSD -ResiliencySettingName Mirror

A parity tier of common HDDs:

New-StorageTier -StoragePoolFriendlyName S2D -FriendlyName "Parity_Tier" -MediaType HDD -ResiliencySettingName Parity

Now you can create a CSV (Cluster Shared Volume):

New-Volume –StoragePoolFriendlyName S2D –FriendlyName CSV001 –PhysicalDiskRedudancy 2 -FileSystem CSVFS_ReFS -Size 200GB

You can display the list of volumes and their redundancy types like this:

Get-VirtualDisk | ft FriendlyName, ResiliencySettingName, PhysicalDiskRedundancy

A new CSV will appear in the Disk Management.

This volume can be used to hos Hyper-V virtual machines or a Scale-out File Server shares.

So, using Storage Spaces Direct of several servers with local disks, you can easily create a software network storage. Due to S2D, fault tolerance of both any pair of disks or servers (4+ node cluster) is provided. S2D cluster automatically starts the process of data reallocating between the rest devices if any faults of disks or servers are detected. In the test environment you can make sure that when you disconnect any two disks, the storage is still available, and VMs on it are running. How to replace a failed drive in the S2D storage I will describe in the next article.