ZFS on Hardware RAID

[jdelisle]

Wouldn't a write-back cache (battery-backed) improve write performance?
Assuming the cache is sufficiently sized for the workload, would it not
have an effect similar to ZIL, meaning low-latency persisted writes?

On Sat, Jan 19, 2019 at 1:41 PM Ira Cooper <ira at wakeful.net> wrote:

> On Sat, Jan 19, 2019 at 11:43 AM Maciej Jan Broniarz <gausus at gausus.net>
> wrote:
>
> > Hi,
> >
> > I want to use ZFS on a hardware-raid array. I have no option of making it
> > JBOD. I know it is best to use ZFS on JBOD, but
> > that possible in that particular case. My question is - how bad of an
> idea
> > is it. I have read very different opinions on that subject, but none of
> > them seems conclusive.
> >
> > Any comments and especially case studies are most welcome.
> > All best,
> >
>
> I've actually built Illumos based ZFS NAS systems based on raid controllers
> and non-raid controllers.
>
> If given the choice, I'd use non-raid controllers, today.  (HBAs)
>
> Why?  Because ZFS handles the "raid" layer well.  The write back cache a
> RAID controller provides, isn't really needed with ZIL, nor desired.
>
> All a raid controller really brings in my experience is added operational
> complexity once you use good enough hardware, mainly good ZIL drives and
> HBAs.
>
> So it isn't really a "bad" idea, more just redundant, expensive and
> inconvenient.  You could have good reasons to do it, like that the RAID
> controller is built into the motherboard etc.
>
> Just make sure your data integrity is #1.  Writeback caches are properly
> battery protected, or not used. etc.
>
> Best,
>
> -Ira
> _______________________________________________
> freebsd-fs at freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-fs
> To unsubscribe, send any mail to "freebsd-fs-unsubscribe at freebsd.org"
>