Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!spool.mu.edu!howland.erols.net!math.ohio-state.edu!jussieu.fr!univ-lyon1.fr!in2p3.fr!swidir.switch.ch!serra.unipi.it!labinfo.iet.unipi.it!luigi From: luigi@labinfo.iet.unipi.it (Luigi Rizzo) Newsgroups: comp.unix.bsd.freebsd.misc Subject: Re: RAID sw? Date: 21 Oct 1996 14:20:17 GMT Organization: Dip. di Ingegneria dell'Informazione, Univ. di Pisa Lines: 57 Distribution: world Message-ID: <54g0r1$trg@serra.unipi.it> References: <chad-3009960810030001@sverige.pengar.com> <325018D3.1131EC4C@lambert.org> <54d5nk$3ou@zwei.siemens.at> NNTP-Posting-Host: labinfo.iet.unipi.it In article <54d5nk$3ou@zwei.siemens.at>, Ingo Molnar <mingo@pc5829.hil.siemens.at> writes: |> [wading through a bunch of unread articles:] |> |> Terry Lambert <terry@lambert.org> wrote: |> |> : RAID calculation in software is extremely expensive, especially |> : for the hamming codes. [...] |> |> why? |> |> RAID 3 Theory of Operation [greatly simplified]: |> ================================================ |> |> 4 disks with real data |> 1 dedicated disk with XOR-ed data. |> basic data unit: 'logical sector' of 4 ordinary sectors This is a 'special' case in which you tolerate 1 failure out of 5 disks. Perhaps Terry had in mind a more general case where you tolerate M failures on N+M disks. In that case, there is a bit more of overhead in computing the remaining M-1 redundant blocks, which are generally computed as a sum or a polynomial over a prime field GF(p). (Although, to be honest, I don't know if there is a RAID specification allowing more than 1 failure). |> a RAID 3 write operation: |> |> a transaction of: |> parallel write to the 4 disks |> and a write to the XOR-ed sector. [the XOR has to be recalculated] |> |> a RAID 3 read operation: |> |> parallel read from all 4 disks actually, it's parallel read from 4 non-failed disks. In case of a disk failure, it also involves XORing data to rebuild the missing block. |> : [...] It is so expensive that no one has really |> : bothered to implement code to do it. |> |> Such RAID boxes are usually a dedicated 486 board with a better |> SCSI adapter [possibly mirrored], using tagged queueing, ensuring |> hot plugging, etc. IMHO, no magic there, really. The key issue is the use of a dedicated cpu vs. the main processor. The same argument is used in the IDE vs. SCSI debate. Luigi -- ==================================================================== Luigi Rizzo Dip. di Ingegneria dell'Informazione email: luigi@iet.unipi.it Universita' di Pisa tel: +39-50-568533 via Diotisalvi 2, 56126 PISA (Italy) fax: +39-50-568522 http://www.iet.unipi.it/~luigi/ ====================================================================