Planning a Storage Spaces deployment

Successful deployment requires careful planning to ensure problems don't arise during or after the deployment process. The following issues should be considered when planning the deployment of Storage Spaces for your organization:

Hardware
Performance
Resiliency
Availability
Management
Scalability

Hardware

The hardware requirements for implementing Storage Spaces define the kinds of physical storage devices and formats that can be used for creating storage pools. These hardware requirements vary depending on whether or not the Failover Clustering feature is being used to provide high availability for your storage pools.

If you are deploying Storage Spaces without Failover Clustering, the supported types of physical storage devices are as follows:

Serial ATA (SATA) hard drives, including solid state drives (SSDs)
Serial Attached SCSI (SAS) hard drives, including SSDs

Note: Although using USB 2.0 hard drives with Storage Spaces is supported, it is not recommended because of the performance limitations of using USB 2.0 to perform simultaneous reads and writes to multiple, connected USB drives.

In addition, these devices can be connected to a server running Windows Server 2012 in the following ways:

Internal connection using SATA or SAS controllers
External connection to a JBOD enclosure of SATA disks
External connection to a SAS storage array

If you are deploying a highly available (HA) Storage Spaces solution, the requirements are as follows:

Two servers running Windows Server 2012 with the Failover Clustering feature installed and configured to use cluster shared volumes (CSVs).
SAS-connected JBODs that comply with the Certified for Windows Server 2012 logo requirements.
You cannot use iSCSI storage devices for an HA Storage Spaces solution.

Note: Storage Spaces cannot be used in conjunction with hardware RAID. Choosing between hardware RAID and Storage Spaces depends on several factors:

Choose Storage Spaces when you want to be able to virtualize and deploy shared storage of data using thin provisioning and save money by using low-cost, commodity hard drives, while ensuring a level of resiliency similar to what RAID can provide.

Choose a hardware RAID solution for a server when you need to ensure data integrity, fault tolerance, and high performance-for example, a RAID 5 or RAID 6 solution implemented using a RAID controller card. You should also choose hardware RAID when you want to ensure data integrity and fault tolerance for a server's boot volume-for example, by configuring the boot volume as a RAID 1 or RAID 10 volume.
If a server has a RAID controller card and multiple hard drives and you want to use Storage Spaces to pool together these drives, you need to configure the RAID card to expose each drive separately to the operating system.

Performance

To ensure optimal performance of Storage Spaces, make sure you consider the following:

Use enterprise-level SAS disks if you can afford them to get optimum performance from your Storage Spaces solution. Make sure you also use top-quality SAS controller cards as well. Most SAS controller cards are backward compatible for connecting to SATA disks, so you can use these if your budget is limited. If top performance is needed and money is no object, consider using SSDs.
Create different storage pools for the different performance needs of your environment, and populate the pools with appropriate kinds of physical disks. For example, a storage pool for a high-speed file server might contain a mix of SSDs and 15,000-RPM SAS disks, while another storage pool used mainly for archiving multimedia presentations could use cheaper 7200-RPM SATA disks.

Resiliency

Provisioning the right type of virtual disk from a storage pool involves the following considerations:

For general-purpose file servers and similar uses, use mirror virtual disks to ensure resiliency against the failure of one or two physical disks in the pool.
For archival data and streaming media, use parity virtual disks. Because of the higher overhead they require when performing random I/O, do not use parity virtual disks for general-purpose file servers.
For temporary storage such as scratch files for data-editing applications, simple virtual disks can be implemented.

Note: You can gain additional resiliency by formatting volumes using the new Resilient File System (ReFS) included in Windows Server 2012. ReFS provides automatic data-integrity maintenance and can be used to add an additional layer of resiliency above and beyond what mirror or parity virtual disks can provide.
Note that some new features of Windows Server 2012-such as data deduplication, which enables greater amounts of data to be stored on disks-are not supported by ReFS-formatted volumes.

Availability

Storage Spaces can be deployed together with the Failover Clustering feature of Windows Server 2012. Such a scenario can be used to deliver continuously available shared storage for your environment by allowing shared storage to transparently fail over to a different node in a clustered file server when necessary. You can also integrate Storage Spaces with CSVs to implement a Scale-Out File Server that incorporates the benefits of using Storage Spaces.

You should keep the following considerations in mind when deploying Storage Spaces together with Failover Clustering:

You need a minimum of three physical disks.
Physical disks must be SAS, not SATA.
Physical disks must support persistent reservations.
Physical disks must pass the failover cluster validation tests.
Only fixed provisioning can be used, not thin provisioning.

Management

After Storage Spaces has been implemented in your environment, you can manage it in the following ways:

Using the File And Storage Services role page in Server Manager
Using cmdlets in the Storage module of Windows PowerShell

For smaller deployments, Server Manager offers a simple way of managing storage pools, virtual disks, volumes, and shares. For larger deployments-for example, when Storage Spaces is used for Hyper-V shared storage in a datacenter or cloud environment-Windows PowerShell provides a way to automate management tasks using scripts.

Note: You can also use WMI directly to manage Storage Spaces using SMAPI-for example, by writing a utility that does this.

Scalability

The scalability of Storage Spaces enables it to be used in various scenarios, ranging from deploying file servers for small businesses to implementing a private-cloud solution for a midsized organization. For example, a single standalone file server using Storage Spaces with a dozen connected physical disks might be used for any of the following purposes:

A general-purpose file server for a small business or department
An archival storage solution for infrequently accessed documents or media files
Virtual machine storage for a Hyper-V host used for test or development

As a second example, you can combine Storage Spaces with Failover Clustering to deploy a highly available, two-node, clustered file server that you can use to provide shared storage for virtual machines running on clustered Hyper-V hosts in your production environment. Such a scenario is possible because the Windows Server 2012 version of Hyper-V supports a new host-clustering option of using Server Message Block (SMB) 3.0 shares for storing your virtual machine files. The file server in this scenario could conceivably scale up to using several hundred disks deployed in rack-mounted SATA or SAS enclosures, and you could use such a file server as the storage for your organization's private-cloud solution.

[Previous] [Contents] [Next]