To further clarify Will's point... Your current setup provides excellent hardware protection, but absolutely no data protection. ZFS provides excellent data protection when it has multiple copies of the data blocks (>1 hardware devices).
Combine the two, provide >1 hardware devices to ZFS, and you have a really nice solution. If you can spare the space, setup your arrays and things to provide exactly 2 identical LUNs to your ZFS box and create your zpool with those in a mirror. The best of all worlds. On Thu, Aug 14, 2008 at 9:41 AM, Will Murnane <[EMAIL PROTECTED]>wrote: > On Thu, Aug 14, 2008 at 07:42, Borys Saulyak <[EMAIL PROTECTED]> > wrote: > > I've got, lets say, 10 disks in the storage. They are currently in RAID5 > configuration and given to my box as one LUN. You suggest to create 10 LUNs > instead, and give them to ZFS, where they will be part of one raidz, right? > > So what sort of protection will I gain by that? What kind of failure will > be eliminated? Sorry, but I cannot catch it... > Suppose that ZFS detects an error in the first case. It can't tell > the storage array "something's wrong, please fix it" (since the > storage array doesn't provide for this with checksums and intelligent > recovery), so all it can do is tell the user "this file is corrupt, > recover it from backups". > > In the second case, ZFS can use the parity or mirrored data to > reconstruct plausible blocks, and then see if they match the checksum. > Once it finds one that matches (which will happen as long as > sufficient parity remains), it can write the corrected data back to > the disk that had junk on it, and report to the user "there were > problems over here, but I fixed them". > > Will > _______________________________________________ > zfs-discuss mailing list > zfs-discuss@opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/zfs-discuss > -- chris -at- microcozm -dot- net === Si Hoc Legere Scis Nimium Eruditionis Habes
_______________________________________________ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
