NTFS is perfectly adequate for most situations. But today's 24/7 data centers require maximum resiliency, and the file system has begun to show its age (it dates all the way back to the 1990s, after all). In an effort to
ReFS is designed to be backward compatible with NTFS, but boasts several new features, the most important of which are aimed at ensuring data integrity.
Checksum for metadata. In ReFS, the checksum is stored separately from the data pages. This allows all manner of disk corruption to be detected.
Of course this isn’t to say that NTFS doesn’t do anything to prevent disk corruption. NTFS has several mechanisms that are designed to maintain data integrity. For instance, NTFS is a transactional file system: Write operations are rolled back if a problem prevents the operation from being completed. Even so, ReFS will take integrity preservation to another level.
Integrity streams. In NTFS, data can be lost in the event of a power failure. Imagine that an existing file is being updated and a power failure occurs during the write operation. When the power comes back on, NTFS can invalidate the incomplete write operation, but at that point the damage is done because the previous version of the file has already been overwritten with new data.
Integrity streams are designed to prevent this type of corruption. When an existing file is updated, the update is written to a different area of the disk. That way, the original data is left intact until a successful write can be confirmed. The confirmation happens through a file level check sum of the data.
Integrity streams are an optional feature that can be enabled or disabled on a server-wide basis, because some applications (particularly database applications) maintain strict control over how data is written to disk. As such, there are times when integrity streams should not be used.
Salvage (also known as Continuous Volume Availability). In NTFS, if corruption occurs on a volume then the solution is usually to take the volume offline and run CHKDSK. The problem with doing so is that the volume is unavailable for use while CHKDSK runs, and in some cases CHKDSK has been known to make the corruption worse.
The ReFS salvage feature -- one of my personal favorite new features -- simply removes the corrupt data from the volume namespace instead of taking the entire volume offline. This helps to prevent the corruption from spreading and affecting good data, and allows the good data to remain available.
More on Windows Server 8 file management
Windows Server 8 data deduplication delivers efficient storage
Improved data mirroring. ReFS allows data to be replicated to multiple disks. Although this can be done with NTFS, ReFS will use mirroring much more intelligently. If the file system detects a checksum error then it will check the replicas to see if it can locate a non-corrupt copy of the data. If a good copy is found then the good data is used to automatically repair the corrupt replicas.
As you can see, ReFS offers a number of improvements over NTFS. Even so, ReFS does have its limitations. For instance, ReFS cannot be used on boot volumes or on removable media. Furthermore, volumes that are currently formatted as NTFS cannot be upgraded to ReFS. These limitations might be an issue for some, since it means that existing servers that may be eventually upgraded to Windows Server 8 will be unlikely to be able to use the new file system. Likewise, servers with a single disk will be unable to use ReFS because of the boot device restriction.
Brien M. Posey, MCSE, is a Microsoft Most Valuable Professional for his work with Windows 2000 Server, Exchange Server and IIS. He has served as CIO for a nationwide chain of hospitals and was once in charge of IT security for Fort Knox. He writes regularly for TechTarget sites.
This was first published in April 2012