Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 1 Feb 2000 19:07:30 +1030
From:      Greg Lehey <grog@lemis.com>
To:        Wilko Bulte <wilko@yedi.iaf.nl>
Cc:        "Justin T. Gibbs" <gibbs@narnia.plutotech.com>, scsi@FreeBSD.ORG
Subject:   Re: hardware vs software stripping
Message-ID:  <20000201190730.R76348@freebie.lemis.com>
In-Reply-To: <20000131191957.A906@yedi.iaf.nl>
References:  <Pine.BSF.4.10.10001301401360.60037-100000@server.b0x.com> <200001311432.HAA32638@narnia.plutotech.com> <20000131191957.A906@yedi.iaf.nl>

next in thread | previous in thread | raw e-mail | index | archive | help
On Monday, 31 January 2000 at 19:19:57 +0100, Wilko Bulte wrote:
> On Mon, Jan 31, 2000 at 07:32:31AM -0700, Justin T. Gibbs wrote:
>> In article <20000131104827.A62824@freebie.lemis.com> you wrote:
>>>
>>> I suppose you mean striping.  RAID-5 doesn't stripe at the byte level,
>>> it stripes at the block level.  RAID-3 stripes at the byte level.
>>
>> I've heard you say this several times, but it is simply not true.
>> RAID-3 is the same as RAID4 without the optimization for partial
>> stripe writes.  In otherwords, in RAID-3, you must read or write
>> a full stripe where RAID-4 adds the ability to perform RMW operations
>> on the parity block of the stripe for sub-stripe updates.  Pluto
>> uses a RAID-3 system in its video server products and it is certainly
>> not striped on a byte level.  (Just as an aside, given the minimum
>> 512 byte sector size of most magnetic media, striping an a per byte
>> basis would be really wasteful).
>
> FWIW the Compaq HSx arrays try hard to distinguish full stripe
> writes on RAID5 and switches to RAID3 behaviour. This is as Justin
> says all on block level (or rather chunk level where a chunk is a
> number of 512byte blocks).

Well, see my other message.  I'd like to call this RAID-4.  But it's
an interesting optimization, and I have a request to implement it in
Vinum.  It should work just as well on RAID-5 as on RAID-4.  The key
is that the write transfers must go exactly across the stripe: then
you can just calculate the parity block and write it without having to
do pre-reads of the area.

> For RAID3 to work well you would like to have synchronised disk
> spindles too (but try to find disks that can do that these days,
> they are not too common. And multiple diskvendors in one RAIDset
> don't mix well with spindle sync). RAID3, I think.., is mostly for
> specialised use (like video, or loading giant datasets on a
> numbercruncher) these days

That's OK, there are plenty of requirements for that, as Justin has
shown.

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-scsi" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000201190730.R76348>