Windows Server 2012 Review: Reliability Page 3
One of the first places you see this theme is in the area of storage. Windows Server 2012 introduces a new concept called Storage Spaces. This is the name for the software-based storage system and in large part replaces, or at least makes unnecessary, RAID for reliable storage.
Storage Spaces manages all disk drives attached to the server using, at the top level, the concept of a Storage Pool. Storage Pools have from one to many physical disks attached and available as a resource from which Volumes can be created.
Individual Volumes may be initialized using one of three layouts: simple, mirror or parity. Simple distributes data across available disks with no redundancy. Mirror makes a copy of data at a "chunk" level and distributes the copies across multiple drives.
Parity uses additional error checking information for each "chunk" and then distributes them across multiple drives. Both mirror and parity require a minimum of two physical disks, while simple will work with a single drive.
Another piece of the reliability puzzle centers on the use of clustering to provide both resilient compute and storage resources. In previous versions of the Windows Server operating system you had to purchase an Enterprise level of the software to get the clustering feature. Now you can get it in the Standard edition. This makes it possible for small-to-medium businesses to take advantage of the power of clustering to provide reliable services that are available all the time.
In the past this key piece of Microsoft's security architecture had to reside on a physical machine. Redundancy was provided by a Backup Domain Controller (BDC) that had to be on a separate physical machine. Now you can create a virtual DC, host it on a cluster and remove both the previous physical DC and BDC machines.
Windows Server 2012 also includes an update to the venerable Server Message Block (SMB) protocol in the form of SMB 3.0. This new release was totally rewritten with reliability and scalability in mind.
SMB 3.0 includes a failover capability that redirects a connected client to a new server in less than three seconds in the case of a failure. To achieve this functionality, both the client application and the server share must be running either Windows 8 or Windows Server 2012. This is primarily due to the low-level drivers that handle the communication and include the ability to detect a failed path through the use of something called a Witness service.
These new features in SMB 3.0 make it possible in Windows Server 2012 to store Hyper-V workloads on SMB shares. When you add in the new Cluster Shared Volumes (CSV) feature, you have high availability compute and storage resources using commodity hardware.
There are some minimal requirements for the commodity hardware, including SAS disks for storage, but nothing exorbitant. It's easy to see how the new reliability features alone will be worth the upgrade to Windows Server 2012 for many customers.
Read more on "Server OS Spotlight" »