> From: zfs-discuss-boun...@opensolaris.org [mailto:zfs-discuss- > boun...@opensolaris.org] On Behalf Of Erik Trimble > > As far as what the resync does: ZFS does "smart" resilvering, in that > it compares what the "good" side of the mirror has against what the > "bad" side has, and only copies the differences over to sync them up. > This is one of ZFS's great strengths, in that most other RAID systems > can't do this.
It's also one of ZFS's great weaknesses. It's a strength as long as not much data has changed, or it was highly sequential in nature, or the drives in the pool have extremely high IOPS (SSD's etc) because then resilvering just the changed parts can be done very quickly. Much quicker than resilvering the whole drive sequentially as a typical hardware raid would do. However, as is often the case, a large percentage of the drive may have changed, in essentially random order. There are many situations where something like 3% of the drive has changed, yet the resilver takes 100% as long as rewriting the entire drive sequentially would have taken. 10% of the drive changed .... ZFS resilver might be 4x slower than sequentially overwriting the entire disk as a hardware raid would have done. Ultimately, your performance depends entirely on your usage patterns, your pool configuration, and type of hardware. To the OP: If you've got one device on one SAN, mirrored to another device on another SAN, you're probably only expecting very brief outages on either SAN. As such, you probably won't see any large percentage of the online SAN change, and when the temporarily failed SAN comes back online, you can probably expect a very fast resilver. _______________________________________________ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
