Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 8 May 2001 09:38:37 +0930
From:      Greg Lehey <grog@lemis.com>
To:        Linh Pham <lplist@closedsrc.org>
Cc:        Bob Greene <rgreene@tclme.org>, "Andrew C. Hornback" <hornback@wireco.net>, Steve Blanzy <sblanzy@aperion.com>, FreeBSD Questions <questions@FreeBSD.ORG>
Subject:   Re: Raid
Message-ID:  <20010508093837.A66817@wantadilla.lemis.com>
In-Reply-To: <Pine.BSF.4.33.0105070917330.97088-100000@q.closedsrc.org>; from lplist@closedsrc.org on Mon, May 07, 2001 at 09:21:33AM -0700
References:  <3AF6CCDB.F1029665@tclme.org> <Pine.BSF.4.33.0105070917330.97088-100000@q.closedsrc.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Monday,  7 May 2001 at  9:21:33 -0700, Linh Pham wrote:
> On 2001-05-07, Bob Greene scribbled:
>
>> Well, you wrote the book; I've only read it.  Is this a vinum specific
>> performance penalty?

I suppose you're talking about the requirement to read before
writing.  No, that's not a Vinum-specific problem.  All RAID-5 systems
must do that.

> I think the performance penalty affects almost every RAID 5 array,
> since you have to update the ECC data on all of the remaining drives
> so that the entire file system is still accessible.

I didn't mention this "penalty", because it's insignificant.  One of
the reasons that Vinum performs better than many "hardware" arrays is
because the processor on a PC is much faster.  Even on the 486/66 on
which I originally developed Vinum, the parity calculations were of
hardly any relevance.

> If I remember correctly... you can still have all your data as long
> as no more than one drive were to fail in a single array at any
> given time.  To unmangle that... if you lost one drive already and
> another one fails... I think your screwed.

Correct.  I mentioned that in my earlier message.  There are
significant performance penalties for recovering data from a degraded
array.  Write performance, on the other hand, is better when the array
is degraded.

Greg
--
Finger grog@lemis.com for PGP public key
See complete headers for address and phone numbers

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010508093837.A66817>