Resilient File System (ReFS) was introduced with Windows Server 2012. Unfortunately, the supported deployment scenarios were very small, and the solution was very clearly still version one. If there’s one thing that people aren’t comfortable deploying an early version of, it’s a file system. There were an immense number of benefits to the new file system, however, and everyone kept waiting for Microsoft to put more emphasis on it and get it into the field. That time has finally come with Windows Server 2016 — it is now the recommended file system for several workloads.
Corruption Detection and Repair
One of ReFS’s claims to fame is its ability to detect and repair file system corruption on the fly. This has been a problem in Windows for years. NTFS can get corruption and you would never know it until you ran a Check Disk command to find it and attempt a repair. Up until very recently, that was an offline operation. As disks got bigger and bigger, that would mean that it would take hours or days to do disk scans on large volumes. On top of that, all of this assumes you even knew there was a problem to kick off the disk scan in the first place. Sometimes corruption would occur and it would linger for months without ever being detected until it had propagated into backups and it was too late to fix it. If a full Check Disk is necessary on ReFS for any reason, it’s a multi-threaded parallel operation that runs substantially faster.
Another feature of ReFS that is very helpful is block cloning specifically around virtualization workloads. This feature allows for metadata-only cloning, which doesn’t have to do any form of actual data movement until a write is issued to it — it’s a copy-on-write function. This can make snapshot creation and deletion nearly instantaneous and greatly improve the performance and minimize the impact of those operations. Furthermore, when creating new thick-provisioned VHDx on ReFS, it also happens nearly instantaneously. ReFS doesn’t have to actually clear existing data on the disk, it has already done that as a background process. With ReFS there is no long zeroing process since it simply allocates the necessary blocks for the new volume. These processes are actually faster on ReFS than through ODX in most situations. Due to these benefits, ReFS is actually the preferred file system for Hyper-V workloads in 2016.
Past Hyper-V, the corruption detection and repair operations now has ReFS as the recommended file system for Exchange 2016 deployments. ReFS is also now supported for SQL 2014 and 2016 (though not necessarily recommended yet).
ReFS is the first major change to Windows file systems in 15 years (NTFS 3.1 with XP). It’s a major change, and it was clearly made to accommodate for modern concerns. All of that being said, there are still some caveats with the new file system, like situations where it may not be the best choice, such as if you need deduplication (not supported) or with certain file workloads (lower performance).
Don’t forget to check back at the other posts in this series!
- What’s New in Windows Server 2016: Hyper-V
- Windows Server Licensing Changes in 2016: Core-Based Licenses
- What’s New in Windows Server 2016: Failover Clustering
- What’s New in Windows Server 2016: Nano Servers
- What’s New in Windows Server 2016: Containers, Part 1
- What’s New in Windows Server 2016: Containers, Part 2
- What’s New in Windows Server 2016: Storage Spaces Direct
- What’s New in Windows Server 2016: Storage Replica
- What’s New in Windows Server 2016: Dedup
You can also watch my webinar touching on a bunch of the new features.