Resiliency for ReFS -Resilient File System

Resiliency

ReFS is able to detect corruptions in a filesystem and repair the corruptions without taking storage offline or making the data unavailable, providing increased reliability and data integrity.

When used in Storage Spaces integration scenarios, ReFSdetects and repairs data corruption automatically, with no downtime. Moreover, if a volume is corrupted and there is no alternative copy of the corrupted data, ReFS removes the corrupted data from the namespace.

During read and write operations, a process named scrubber occasionally scans the volume and automatically corrects the errors if the corrupt data is found. 

ReFS

ReFS (Resilient File System) is a new file system that is available in Windows Server 2012, Windows Server 2012 R2, and Windows Server 2016. ReFS is appropriate for very large workloads where you need to maximize data availability and integrity and ensure that the file system is resilient to corruption.

The ReFS file system is suitable for hosting specific types of workloads such as virtual machines and SQL Server data, because it includes the following features that improve upon NTFS:

• Integrity. ReFS uses checksums for both metadata and file data. This means that ReFS can detect data corruption.

• Storage space integration. When integrated with storage spaces that are configured with mirror or parity options, ReFS has the ability to automatically detect and repair corruption using a secondary or tertiary copy of data stored by Storage Spaces. Repair occurs without downtime.

• Proactive error correction. ReFS includes a data integrity scanner that scans the volume to identify latent corruption and proactively repair corrupt data.

• Scalability. ReFS is specifically designed to support data sets in the millions of terabytes range.

• Advanced VM operations. ReFS includes functionality specifically to support virtual machine operations. Block cloning accelerates copy operations, which accelerate VM checkpoint merges. Sparse VDL allows ReFS to substantially reduce the amount of time required to create very large fixed sized virtual hard disks.