Date: Wed, 13 May 2009 07:51:04 -0700 From: Kip Macy <kmacy@freebsd.org> To: Ivan Voras <ivoras@freebsd.org> Cc: freebsd-fs@freebsd.org Subject: Re: UFS2 and SSDs Message-ID: <3c1674c90905130751s60757be2t8039965b71c75467@mail.gmail.com> In-Reply-To: <guelk2$jab$1@ger.gmane.org> References: <49FC1BD0.4030306@T-Online.de> <guecop$jtd$1@ger.gmane.org> <3c1674c90905130529r70589318tf57198d24cf2bd57@mail.gmail.com> <20090513135335.GA42884@voi.aagh.net> <guelk2$jab$1@ger.gmane.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, May 13, 2009 at 7:31 AM, Ivan Voras <ivoras@freebsd.org> wrote:
> Thomas Hurst wrote:
>> * Kip Macy (kmacy@freebsd.org) wrote:
>>
>>> I accidentally bought a camera-grade SSD. Random write performance
>>> with UFS made it unusable. I ended up converting /usr to ZFS - since
>>> which time I've been very happy with performance.
>>
>> Did you try gjournal on it? =A0SSD's should do better with sequential
>> journal writes.
>
> My guess is that it won't matter - the issue is "small writes" not
> "sequential writes". Gjournal will issue writes as it receives them - if
> it receives a bunch of small ones, it will pass them on in the same
> form, only sequential (the drive will still see a bunch of small
> writes). This works well for mechanical drives because of rotational
> properties but does nothing to SSDs.
>
> ZFS OTOH does a great deal of buffering.
The benefits come from write-allocate - writes always end up being
some multiple of erase blocks. With FFS the drive constantly has to GC
partial blocks.
-Kip
--=20
When bad men combine, the good must associate; else they will fall one
by one, an unpitied sacrifice in a contemptible struggle.
Edmund Burke
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3c1674c90905130751s60757be2t8039965b71c75467>