Nope, data is still checksummed on read, and as long as you have some
kind of redundancy in your pool setup zfs will rebuild those bad
blocks it finds with the good replicas.

You lose some capacity and speed if you have HW raid and then raidz at
the zfs layer, but it works fine. Best is to have jbods so zfs can
"see" the disks raw, but I have used it with the dell perc hw raid
where I exported many single disk luns and then made a zpool over the
top of those.

zfs is great in many ways. Definitely worth a look.

On Feb 17, 2010, at 8:01 PM, Jean-Francois Panisset wrote:

I'm curious (and keep in mind that I have zero practical experience with ZFS), one of the most attractive features of ZFS to me is the multi-level checksuming that is supposed to help preserve data integrity, and is a lot more sophisticated than what hardware RAID controllers typically do. So if you build a ZFS storage pool on top of RAIDed LUNs instead of bare drives in a JBOD arrangement, aren't you going against the philosophy of ZFS, or at the very least guilty of "belt and suspenders" engineering?

JF

On Wed, Feb 17, 2010 at 2:31 PM, matxdotca matxdotca@gmail.com
wrote: > > On 2010-02-17, at 1:55 PM, Jean-Francois Panisset wrote: > >> Or get a non-RAID HBA, install Nexenta OS on the server and enjoy >> ZFS goodness > > Doing just that with a bunch of SuperMicro RAID boxen and a Nexenta
> frontend. Great backup snapshot system for our Xsan (the main
> storage production setup). > > Mat X > _ StudioSysAdmins-Discuss mailing list StudioSysAdmins-Discuss@mailman.studiosysadmins.com http://mailman.studiosysadmins.com/mailman/listinfo/studiosysadmins-discuss

---

StudioSysAdmins-Discuss mailing list StudioSysAdmins-Discuss@mailman.studiosysadmins.com http://mailman.studiosysadmins.com/mailman/listinfo/studiosysadmins-discuss