I've got three Linux software RAIDs set up: md0 with 900 GB of M.2 NVMe in RAID 10, md1 with 14 TB of U.2 NVMe in RAID 10, and md2 with 37 TB of SATA SSDs in RAID 6. I've set the sync/rebuild speed for all arrays to 1 GB/s. During my monthly checks, md0 and md1 take about the expected times (15 minutes for md0 and 4 hours for md1), but md2 finishes surprisingly quickly—in just 2 hours, even though it's nearly three times the size of md1. I'm curious about the cause of this discrepancy. Does RAID 6 have an advantage in efficiency for checks compared to RAID 10? Is it true that only allocated space needs to be validated in RAID 6?
3 Answers
Honestly, there are many factors that could affect your disk array speeds. The filesystem types, size of files, stripe widths, and how the disks are connected all matter. RAID 10 requires a lot more checks due to its mirroring, while RAID 6 is more straightforward, particularly if it has fewer disks involved. You mention your system is well-optimized, but be mindful of how everything is set up—PCIe lanes and HBA setups can dramatically influence performance.
RAID 6 is generally faster during validation because it only reads and checks the parity data. This is a single-read and compute operation, which is more efficient than RAID 10, where it has to check each block against its mirror—essentially doubling the workload. Plus, RAID 6 checks tend to be more sequential, which is usually quicker than the random access pattern seen in RAID 10 validations.
There might be some cache factors at play, too. If the RAID 6 setup benefits from superior caching, that could speed things up significantly. A good cache can help a lot with read speeds!
Related Questions
Can't Load PhpMyadmin On After Server Update
Redirect www to non-www in Apache Conf
How To Check If Your SSL Cert Is SHA 1
Windows TrackPad Gestures