Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 29 Aug 2015 15:55:04 +0100
From:      Gary Palmer <gpalmer@freebsd.org>
To:        Quartz <quartz@sneakertech.com>
Cc:        freebsd-fs@freebsd.org
Subject:   Re: Options for zfs inside a VM backed by zfs on the host
Message-ID:  <20150829145504.GA99821@in-addr.com>
In-Reply-To: <55DFA786.8090809@sneakertech.com>
References:  <CALd%2BdcfJ%2BT-f5gk_pim39BSF7nhBqHC3ab7dXgW8fH43VvvhvA@mail.gmail.com> <20150827061044.GA10221@blazingdot.com> <20150827062015.GA10272@blazingdot.com> <1a6745e27d184bb99eca7fdbdc90c8b5@SERVER.ad.usd-group.com> <55DF46F5.4070406@redbarn.org> <453A5A6F-E347-41AE-8CBC-9E0F4DA49D38@ccsys.com> <CALcn87yArcBs0ybrZBBxaxDU0y6s=wM8di0RmaSCJCgOjUHq9w@mail.gmail.com> <55DFA786.8090809@sneakertech.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Aug 27, 2015 at 08:12:54PM -0400, Quartz wrote:
> > I am right now exploring the question: are SSD ZILs necessary in an all SSD
> > pool?
> 
> Something mentioned in another recent thread on this list (or maybe it 
> was -questions?) was that yes, you really should consider a separate ZIL 
> if you're using primarily SSDs. Without a separate disk, log writes have 
> to steal blocks from the pool itself which then have to be deleted 
> afterwards to let go of the space. Besides causing excess file 
> fragmentation, the write-delete cycle doesn't play well with SSDs and 
> trim and can seriously hamper performance. With a dedicated disk, it 
> writes and then just leaves it there, only overwriting later if necessary.

Presumably they're only necessary if you're dealing with sync writes?  If
the vast majority of your workload is async writes, does a separate ZIL SSD
still help?

And I still am curious why ZFS has no stats for letting you measure sync
vs async writes

Gary



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20150829145504.GA99821>