All of lore.kernel.org
 help / color / mirror / Atom feed
From: Stan Hoeppner <stan@hardwarefreak.com>
To: "Nuno Magalhães" <nunomagalhaes@eu.ipp.pt>, linux-raid@vger.kernel.org
Subject: Re: RAID newbie, 1 vs 5, chunk sizes
Date: Sat, 14 Jun 2014 16:56:08 -0500	[thread overview]
Message-ID: <539CC4F8.90905@hardwarefreak.com> (raw)
In-Reply-To: <CADqA9ua4xuJM3J6mQY7jXrt-5f6Rsv3aFaxyiqL1e80t70R6Rg@mail.gmail.com>

On 6/14/2014 4:01 PM, Nuno Magalhães wrote:
...
> This is a desktop system, nothing
> mission-critical, but i'd like it to be reasonably tailored to the
> hardware and intended usage, and have questions about default values.

[snip]

Everything begins and ends with the workload.  The proper chunk and
stripe sizes are dictated by the write patterns of your workload, i.e.
the mix of applications running on the system, not the type/size of
disks you have.

If you're writing lots of small files then you'll typically want a small
chunk size.  If you use a large chunk size with small files you'll get a
hot spot on one or more drives in the array due to filesystem alignment.
 Hot spot means one disk sees lots of activity while the other(s) sit
idle.  This defeats the purpose of striping.

If you're working predominantly with large media files you'll want a
large chunk/stripe for data transfer efficiency from the platters with
the fewest seeks.  If you have a mixed workload with small and large
files, it's best to use a small chunk size, such as 32 or 64KB.


All of that stated, your filesystem setup, specifically write barrier
support and journal mode, along with those slow disk spindles, will be a
much more significant impediment to performance than your chunk size.

Cheers,

Stan
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

  reply	other threads:[~2014-06-14 21:56 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-06-14 21:01 RAID newbie, 1 vs 5, chunk sizes Nuno Magalhães
2014-06-14 21:56 ` Stan Hoeppner [this message]
2014-06-15 11:50   ` Nuno Magalhães
     [not found]     ` <pmrj5x4rq3qbl6unnu7guho9.1402849407134@email.android.com>
2014-06-15 17:40       ` Nuno Magalhães
2014-06-16  2:23 ` Phil Turmel
2014-06-16 12:36   ` Nuno Magalhães
2014-06-16 13:19     ` Mark Knecht
2014-06-16 14:28     ` Phil Turmel
2014-06-16 16:14       ` Nuno Magalhães
2014-06-16 20:13         ` Phil Turmel
2014-06-16 21:02           ` Mark Knecht
2014-06-17  0:41         ` Nuno Magalhães
2014-06-17  0:47           ` Brad Campbell
2014-06-17  9:09             ` Nuno Magalhães
2014-06-17 19:50               ` Stan Hoeppner
2014-06-18  2:55               ` Brad Campbell
2014-06-18  7:33     ` Wilson, Jonathan
2014-06-18 10:19     ` Dag Nygren
2014-06-18 11:58       ` Mathias Burén
2014-06-18 12:16         ` Brad Campbell
2014-06-18 17:34         ` Roman Mamedov
2014-06-19  6:45           ` Wilson, Jonathan
2014-06-29 13:22             ` Nuno Magalhães

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=539CC4F8.90905@hardwarefreak.com \
    --to=stan@hardwarefreak.com \
    --cc=linux-raid@vger.kernel.org \
    --cc=nunomagalhaes@eu.ipp.pt \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.