draid · shipping · also OpenZFS dRAID, distributed RAID

dRAID (OpenZFS)

OpenZFS dRAID is a RAID-Z variant with fixed-width redundancy groups distributed across child devices and integrated distributed-spare capacity.

OpenZFS 2.3 spreads reconstruction work across children instead of funneling writes into one spare disk. Its fixed-width layout permits sequential resilvering but pads small blocks; eight data columns with 4 KiB sectors have a 32 KiB minimum allocation. Configured distributed spares serve only their own dRAID vdev. [1]

Sequential reconstruction restores redundancy without verifying checksums and is followed by a scrub. That is a separate verification stage, not a promise of file-level access beyond parity tolerance. D-GRAID is a different research design. [2] [3]

Relations

related
RAIDZ, Parity declustering, Distributed spare, Resilver, D-GRAID

Limitations

  • Fixed-width padding can reduce usable capacity and effective compression for small blocks. [1]
  • Distributed spare capacity is a recovery resource, not extra parity. [1]

Citations

  1. OpenZFS 2.3 zpoolconcepts(7): dRAID virtual devices and hot spares. (accessed 2026-11-09)
  2. OpenZFS 2.3 zpool-replace(8): sequential reconstruction and subsequent scrub. (accessed 2026-11-09)
  3. Sivathanu, Prabhakaran, A. C. Arpaci-Dusseau and R. H. Arpaci-Dusseau. Improving Storage System Availability with D-GRAID. FAST 2004, sections 3, 5-7 and 9. (accessed 2026-11-09)

Further reading

  1. Hesela: partial file availability, dRAID and recovery milestones (accessed 2026-11-09)

JSON: /data/terms/draid.json

Published by Hesela team · Updated

Report a correction · Editorial standards · Reuse guidance

Cite this page
Hesela team. dRAID (OpenZFS). Hesela. 2026-11-09. https://hesela.dev/terms/draid/