XRAID brings declustered parity RAID to zoned XFS in first release
Christoph Hellwig has published an initial public release of XRAID, a parity RAID layer integrated into XFS. It targets zoned devices such as SMR drives and claims write speeds far above mdraid.

Key takeaways
- Christoph Hellwig has released XRAID, a declustered parity RAID layer integrated into XFS.
- It supports parity RAID on zoned devices such as SMR hard drives and works on conventional drives and SSDs too.
- Hellwig claims writes are up to 20 times faster than mdraid, even without SSD-specific optimisation.
- The code is not ready for upstream submission, but the developers are seeking wider testing and feedback.
In this article
First public release of XRAID
Christoph Hellwig has announced the initial public release of XRAID, a declustered parity RAID implementation that is integrated into the XFS file system. According to LWN.net, the announcement was published on 13 November 2026.
The project has two main goals. The first is to support parity RAID on zoned devices, such as shingled magnetic recording (SMR) hard drives, which write data in sequential zones. The second is to offer a fast RAID implementation without the inherent data integrity problems the project says plague existing approaches.
Work on XRAID has been going on for a year and a half. Hellwig says the project is not yet anywhere near ready for submission to the upstream kernel, but the developers now want feedback and testing from a wider audience.
Performance claims compared with mdraid

Although SSD optimisation has not been a focus, XRAID also runs on conventional hard drives and SSDs. Hellwig reports that even without SSD-specific tuning, writes are up to 20 times faster than mdraid, the RAID implementation already present in the Linux kernel.
Readers should treat the performance figure as a claim from the developer rather than an independently verified benchmark. No detailed test methodology accompanies the announcement.
The declustered design is what makes the zoned support possible. Parity information is spread across drives in a way that suits the sequential write model of zoned devices, which is where the integration with zoned XFS matters.
Parity levels and planned expansion
The current implementation supports zero to 2 parities. Hellwig says the team has run experiments with the SnapRAID parity library to enable 3 or more parities.
That extended support is not included in the current version. However, the developers say support for additional parities is planned, and the code has been prepared for it.
This gives XRAID a path towards configurations with more redundancy than a standard double-parity setup, though the timeline for that work is not stated.
Tools and how to try it
The xfsprogs branch includes mkfs support for creating XRAID file systems, plus a new xraid utility for managing devices and inspecting the resulting file systems.
The developers have also extended Zonar, a visualisation tool for zoned file systems, so it can show how each logical zone maps onto physical zones on the underlying devices.
Hellwig gave a presentation on the topic at FOSSY in August. Video of the talk does not appear to be available yet, but his slides can be found online for those who want more technical background.
Why it matters
Parity RAID on zoned devices has been a gap in the Linux storage stack. SMR drives offer high capacity at low cost, but their sequential write model makes traditional RAID approaches problematic. A design built around zoning, combined with XFS support for zoned devices, could make these drives practical for data storage arrays.
The claimed speed advantage over mdraid is also notable, though the project is clearly marked as early work. Interested users can test it now, but anyone relying on it for important data should wait until the code matures and receives wider review.
Sources
Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.



