Probably a Sneak Peek, unless there’s a stealthy mountain involved.
Glad to see some of the lessons from cluster scale filesystems slowly leaking into open source data storage products.
I’m struggling a bit to understand how this is different from RAID-Z2 and RAID-Z3. In these modes, the VDEV can tolerate two or three devices failing and, when that happens, can either run in the degraded state or can have a spare onlined to resilver. You don’t get back up to the full redundancy level until you’ve finished this process. It sounds as if there’s something different between the load on the disks from resilvering RAID-Z2 and dRAID, but I can’t quite figure out what from the article.