Article View: comp.arch.storage
Article #469Re: RAID 1 or 5 ?
From: Bill Todd
Date: Sat, 02 Oct 1999 00:00
Date: Sat, 02 Oct 1999 00:00
83 lines
3881 bytes
3881 bytes
Multiple issues going on here. Malcolm's answer was accurate, Ron's was either confused or confusing. Don't muddle things by comparing arrays with the same number of disks: you need however much (non-parity) disk space you need, period, so compare arrays with equal amounts of (non-parity) space. In that case, RAID-5 1) is cheaper (n + 1 disks vs. 2n disks for RAID-1), 2) has single-threaded Read performance similar to RAID-1 (assuming RAID-1 uses striped mirroring when n > 1), 3) has multi-threaded Read performance inferior to RAID-1 (RAID-1 can distribute the Reads over twice as many disks), 4) has significantly slower write performance than RAID-1 (unless you're doing stripe-aligned full-stripe writes, in which case performance is about equal), 5) is less robust than RAID-1 if two disks fail simultaneously (when n > 1, two disk failures destroy RAID-1 data only if they hit a mirror pair), and 6) suffers more performance degradation during rebuild of a failed disk than RAID-1 (the entire array is involved in the rebuild activity, not just one disk). RAID-1 also has the advantage of somewhat more flexible configurability - e.g., if you concatenate (rather than stripe) the disk set, it may be easier to expand it incrementally (given a file system that can deal with that), and if you can find a controller that supports it you can choose to use a 'chained declustered' layout that lessens performance degradation during replacement of a failed disk (though at the expense of the added robustness described in -5- above). Hope that helps, - bill Shaman <shaman@ihug.co.nz> wrote in message news:37F548BD.AFB2D67@ihug.co.nz... > Hi > > Anyway back to the topic (RAid 1 vs 5; write performance and overall > performance for a file and print server with a hardware RAID card.) > > I was aware of the need to calculate parity during the write in RAID 5 but I > thought the ability to write a small peice of the whole file to each disk (in > RAID 5) would be more significant timewise. > > Does RAID 5 ever get better than RAID 1, as you add disks? > > ie RAID 1 vs a 8 Disk RAID 5 > is RAID 1 still better? > > David > > > > Jim Pappe wrote: > > > with Raid 5 plus hot spare - 2 drive can fail .. > > www.raidexpert.com > > > > Mark Landin <m555@earthlink.net> wrote in message > > news:37e660eb.12904299@news.earthlink.net... > > > On Fri, 17 Sep 1999 07:54:19 +0100, "Steven Perez" > > > <steve@uninet.demon.co.uk> wrote: > > > > > > >Kenneth: > > > > > > > >Go with RAID 1/0 Stripe, and mirror. This will give the very best > > > >performance, and availability, although it will give the least amount of > > > >storage. > > > > > > How is RAID 0/1 more "available" than RAID 5? > > > > > > >
Technical Details
Message-ID:
<7t5grt$1t9$1@pyrite.mv.net>Path:
arc1-ca.pugleaf.net!unknown!not-for-mailReferences:
<7nl85m$3nn$1@newton.pacific.net.sg> <37E1425E.9F4D1F3B@t-online.de> <937569107.3977.0.nnrp-11.9e98b72c@news.demon.co.uk> <37e660eb.12904299@news.earthlink.net> <7sehd5$80a@journal.concentric.net> <37F548BD.AFB2D67@ihug.co.nz>Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: f4d7c,813a29bed80cb2f6 X-Google-Attributes: gidf4d7c,public X-Google-Thread: 10e1c8,813a29bed80cb2f6 X-Google-Attributes: gid10e1c8,public X-Google-Thread: 1144af,813a29bed80cb2f6 X-Google-Attributes: gid1144af,public From: "Bill Todd" <billtodd@foo.mv.com> Subject: Re: RAID 1 or 5 ? Date: 1999/10/02 Message-ID: <7t5grt$1t9$1@pyrite.mv.net> X-Deja-AN: 532061511 References: <7nl85m$3nn$1@newton.pacific.net.sg> <37E1425E.9F4D1F3B@t-online.de> <937569107.3977.0.nnrp-11.9e98b72c@news.demon.co.uk> <37e660eb.12904299@news.earthlink.net> <7sehd5$80a@journal.concentric.net> <37F548BD.AFB2D67@ihug.co.nz> X-Priority: 3 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300 X-Complaints-To: abuse@mv.com X-Trace: pyrite.mv.net 938886845 1961 199.125.99.136 (2 Oct 1999 17:54:05 GMT) Organization: Why won't Outlook let me leave this blank? X-MSMail-Priority: Normal NNTP-Posting-Date: 2 Oct 1999 17:54:05 GMT Newsgroups: comp.arch.storage,comp.sys.ibm.pc.hardware.storage,fido.ger.storage