Obviously, maintaining the availability of Server-based data is one of the primary objectives of any Server Administrator. With the “Data Verification and Auto Correction” process provided by ReFS we’re able to support that availability goal. While the self-healing aspects of such a file system existed in the Server 2008 versions of the operating system, ReFS takes that support capability to a level of higher resiliency. As described above, ReFS can auto-detect data corruption and then “self-heal.” This means that the file system can perform any corrections and repairs necessary without needing to take the file system offline.
Most systems administrators routinely perform disk checking as part of a maintenance script or in an “out of band” process from the GUI in order to determine file system errors and repair bad disk sectors. With an ReFS formatted volume, one cannot even run the CHKDSK utility so performing a disk integrity check is integrated as part of ReFS as a file system.
For data availability the majority of System Administrators have traditionally relied on some type of RAID solution. But in the unusual circumstance where a RAID volume becomes corrupted to the point where a full-blown data recovery operation is necessary, the amount of downtime is a quantifiable cost factor to the business process and operation. Having a RAID 1 volume become corrupt or even losing two or more volumes in a RAID 5 solution is a rare-case scenario, but it does happen.
It is in these instances that ReFS can save the day. ReFS has salvage capability whereby using data verification and auto correction corrupt data is removed from its location on an accessible “in production” volume. This means that a live, online volume, in a way, self-heals at the file level. The end result of this operation is that in the event of data corruption of an accessible file, the corrupt, or “bad” data, is deleted from the file system and volume while the remaining files on that volume continue to be accessible and online. This is a transparent operation and occurs on average in less than one second.
More specifically, the highest level of data protection occurs on fault tolerant volumes (some level of RAID). So, when a process attempts to perform a read operation and throws an error on a mirrored volume, the file system driver, as implemented through Storage Spaces, reads the data file from both volumes. This is where ReFS works in conjunction with Storage Spaces and determines which of the file volumes is corrupt, selects the uncorrupted file volume, and auto repairs the corrupt one. If a write operation is taking place, the file system driver directs the device driver to write the data to a location on the volume that is viable or not corrupt.
When files become corrupt on a single non-fault tolerant drive, the file system drivers are unable to “grant a handle” to the object manager. In the absence of ReFS, the System Administrator is forced into some type of time-consuming recovery process that involves taking the system offline. The norm at present is that the file system cannot open or delete a corrupt file, making it difficult for an administrator to respond without taking the system offline. But because ReFS can still function around the corrupt data, the administrator is able to recover that file from a copy or backup, or have the application re-create it without taking the file system offline. This key innovation ensures that we do not need to run an expensive offline disk checking and correcting tool, and it allows for very large data volumes to be deployed without risking costly offline periods due to corruption.
Designed in conjunction with ReFS, Storage Spaces is referred to as a virtualization capability deployable through Server 2012. The overriding objective of Storage Spaces is to reduce the cost of supporting and maintaining the high availability of physically hosted and virtual machine-based data storage. Through the creation of storage pools, a common storage space is made available to the Server 2012 Operating System that exists on any available and compatible physical storage device. Most Server Administrators are familiar with the concept and implementation of storage pools, the advantage of Storage Spaces as implemented through the Server 2012 Operating System is two-fold.
First, by taking a collection of physical drives, commonly referred to as Just a Bunch of Drives (JBOD), one is able to create one or more logical Storage Spaces and scale each of those spaces into a single logical volume. This volume is extensible as the System Administrator is able to add, or aggregate, to the storage space as demand requires. This dramatically cuts down on the number of individual storage devices the administrator has to support and maintain. Storage Spaces supports any commonly used physical storage device, including USB, Serial ATA (SATA), or Serial attached SCSI (SAS). Hence, the reference to JBOD.
The second most apparent benefit in using Storage Spaces is the management interface of these storage spaces. One can choose to manage the Storage Space through the File and Storage GUI or through Powershell (highly recommended here). In either management approach, the System Administrator has the capability to view all drives as members of their respective pool and retrieve functioning status information. From this perspective the Administrator has at his or her disposal the ability to isolate failed physical drives, and remove them, for example, or simply extend the Storage Space by adding new drive to the pool. Storage Spaces also supports common fault tolerant technologies such as data Striping, Mirroring, and Striping with Parity. As stated previously, using ReFS in conjunction with Storage Spaces provides the most robust and fault-tolerant storage solution.
ReFS, the Natural Evolution and Improvement of NTFS
As with its tried and true predecessor, NTFS, ReFS delivers significant enhancements to the Server 2012 File System. It is a natural evolution to the constant and inherent changes in Information Technology. The integrity and availability of data storage are paramount and, given the rate of technological change in this industry, it is only natural that the file system technologies that support such storage evolve to keep pace. Undeniable changes such as virtualization, data accumulation, and data sprawl simply mandate that ReFS-like technologies emerge and become de facto standards.
ReFS delivers features such as metadata consistency and its modifications, the use of checksums to enable the detection of data corruption and subsequent repair without user intervention. In addition, Storage Spaces provide a flexible and extensible method to data storage that may well prove as efficient and more cost-effective than proprietary hardware vendor solutions. By taking advantage of ReFS and the newly released Windows Server 2012, there is a distinct value-add for the data center in particular. This is a very compelling feature of Server 2012 and offers System Administrators the ability to further standardize their approach to managing the data center. What was once capability found only in the purchase of a third-party solution, is now a flexible and scalable feature built into the operating system.