From: Bill Davidsen <davidsen@tmr.com>
To: Daniel Korstad <dan@korstad.net>
Cc: gmitch64@yahoo.com, linux-raid@vger.kernel.org
Subject: Re: Software based SATA RAID-5 expandable arrays?
Date: Sat, 14 Jul 2007 11:49:10 -0400 [thread overview]
Message-ID: <4698F076.7080507@tmr.com> (raw)
In-Reply-To: <920282231.1184166214119.JavaMail.root@gateway.korstad.net>
Daniel Korstad wrote:
>
> That was true up to kernel 2.6.21 and 2.6 mdadm where support for RAID 6 reshape arrived.
>
> I have reshaped (added additional drives) to my RAID 6 twice now with no problems in the past few months.
>
> You mentioned that as the only disadvantage. There are other things to consider. The overhead for parity of course. You can't have a RAID 6 with only three drives unless you build it with a missing drive and run degraded. Also (my opinion) it might not worth the overhead with only 4 drives, unless you plan to reshape (add drives) down the road. When you have an array with several drives, than it is more advantages as the percentage of disk space lost to parity goes down [((2/N)*100) where N is the number of drives in the array] so your storage efficiency increases ((Number of Drives -2)/Number of Drives). And with more drives the statistics of getting hit with a bit error after you lose a drive and you are trying to rebuild increases.
>
> Also, there is a very slight performance drop for write speeds on RAID6 since you are calculating p and q parity.
>
>
I would expect (and see) a fairly substantial drop in write performance.
With RAID-5 only the parity needs to be read on a data change, and the
old data chunk. Then several XORs are done and the new data and new
parity written. With RAID-6, I believe that all the data in the stripe
need be read for calculating the q parity.
> But for what I use my system for, family digital photos, file storage and media server I mostly read data and not bothered with slight performance hit in write.
>
> I have been using RAID6 with 10 disk for over a year and it has saved me at least once.
>
> As far as converting the RAID6 to RAID5 or RAID4... Never had a need to do this, but no probably not.
Agree, for many things the write performance is not an issue, while the
reliability is. Backups are still desirable, of course.
--
bill davidsen <davidsen@tmr.com>
CTO TMR Associates, Inc
Doing interesting things with small computers since 1979
next prev parent reply other threads:[~2007-07-14 15:49 UTC|newest]
Thread overview: 40+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-07-11 15:03 Software based SATA RAID-5 expandable arrays? Daniel Korstad
2007-07-14 15:49 ` Bill Davidsen [this message]
-- strict thread matches above, loose matches on Subject: below --
2007-08-01 16:46 Michael
2007-08-01 17:13 ` David Greaves
2007-07-16 17:34 Michael
2007-07-16 19:29 ` Daniel Korstad
[not found] <1914474980.1184590115562.JavaMail.root@gateway.korstad.net>
2007-07-16 14:23 ` Daniel Korstad
2007-07-13 16:48 Michael
2007-07-13 18:18 ` Bill Davidsen
2007-07-13 18:23 ` Daniel Korstad
2007-07-12 22:42 Michael
2007-07-13 3:54 ` Bill Davidsen
2007-07-13 15:22 ` Daniel Korstad
2007-07-11 20:12 jahammonds prost
2007-07-11 20:08 Michael
2007-07-11 23:29 ` Nix
2007-07-11 17:26 jahammonds prost
2007-07-11 19:13 ` Daniel Korstad
2007-07-11 19:26 ` Daniel Korstad
2007-07-10 21:58 jahammonds prost
[not found] <944875.74303.qm@web54106.mail.re2.yahoo.com>
2007-07-09 19:31 ` Daniel Korstad
2007-07-11 14:21 ` Bill Davidsen
2007-06-22 15:12 jahammonds prost
2007-06-23 4:16 ` Brad Campbell
2007-06-21 10:40 Michael
2007-06-21 9:22 Michael
2007-06-21 20:16 ` Richard Scobie
2007-06-19 14:42 Michael
2007-06-19 21:30 ` Nix
2007-06-19 13:43 Michael
2007-06-19 14:23 ` Robin Hill
2007-06-19 18:49 ` Daniel Korstad
2007-06-19 13:08 Michael
2007-06-18 12:46 Daniel Korstad
2007-06-17 22:16 greenjelly
2007-06-17 22:23 ` Justin Piszcz
2007-06-18 21:14 ` Dexter Filmore
2007-06-19 8:35 ` David Greaves
2007-06-19 9:14 ` Dexter Filmore
2007-06-20 20:52 ` Brad Campbell
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4698F076.7080507@tmr.com \
--to=davidsen@tmr.com \
--cc=dan@korstad.net \
--cc=gmitch64@yahoo.com \
--cc=linux-raid@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).