From mboxrd@z Thu Jan 1 00:00:00 1970 MIME-Version: 1.0 In-Reply-To: References: <20110810220404.D6021B827@mail.bitblocks.com> <3257bf7eda3d7965220be7315b21cb86@coraid.com> <7d25ae781f59985f949b025bbc0324bb@coraid.com> Date: Thu, 11 Aug 2011 11:45:49 -0700 Message-ID: From: David Leimbach To: Fans of the OS Plan 9 from Bell Labs <9fans@9fans.net> Content-Type: multipart/alternative; boundary=0015174beec49b1e0404aa3f3411 Subject: Re: [9fans] simple venti demo: Topicbox-Message-UUID: 0f84a780-ead7-11e9-9d60-3106f5b1d025 --0015174beec49b1e0404aa3f3411 Content-Type: text/plain; charset=ISO-8859-1 On Thu, Aug 11, 2011 at 10:37 AM, ron minnich wrote: > On Thu, Aug 11, 2011 at 10:28 AM, erik quanstrom > wrote: > this is the same > > dillema any non content-addressed disk has. performance > > vs. safety. and of course one size doesn't fit all, so there are knobs > in > > most disks to turn off write caching. > > it's not as obvious a tradeoff as it seems. > > Anyway, I'm more interested in hearing from people who do something > with the code. > > ron > > Unless before you write to venti, you store stuff durably locally, and have a way to replay lost data, you can't be sure the data in venti is really stored, when venti ACKs your request to store data, it's nice to know it's durably there. So yes, it will potentially perform worse, but you're trading off actual durability for performance. One could go on tangents for crash-only designs, fault tolerance etc... but that'd be digging the hole deeper. Dave --0015174beec49b1e0404aa3f3411 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable

On Thu, Aug 11, 2011 at 10:37 AM, ron mi= nnich <rminnich@= gmail.com> wrote:
On Thu, Aug 11, 2011 at 10:28 AM, erik quanstrom
<quanstro@= labs.coraid.com> wrote:
=A0=A0this is the same
> dillema any non content-addressed disk has. =A0performance
> vs. safety. =A0and of course one size doesn't fit all, so there ar= e knobs in
> most disks to turn off write caching.

it's not as obvious a tradeoff as it seems.

Anyway, I'm more interested in hearing from people who do something
with the code.

ron


Unless before you write to vent= i, you store stuff durably locally, and have a way to replay lost data, you= can't be sure the data in venti is really stored, when venti ACKs your= request to store data, it's nice to know it's durably there.

So yes, it will potentially perform worse, but you'= re trading off actual durability for performance.

= One could go on tangents for crash-only designs, fault tolerance etc... but= that'd be digging the hole deeper.

Dave

--0015174beec49b1e0404aa3f3411--