mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Kurt Garloff <garloff@suse.de>
To: Jens Axboe <axboe@suse.de>
Cc: Linux kernel list <linux-kernel@vger.kernel.org>
Subject: Priorities for I/O
Date: Mon, 21 Oct 2002 09:26:29 +0200	[thread overview]
Message-ID: <20021021072629.GD6630@nbkurt.casa-etp.nl> (raw)

[-- Attachment #1: Type: text/plain, Size: 1432 bytes --]

Hi Jens,

one of the shortcomings of Linux process priority system is that it only
affects CPU resources (and even those are not affected strongly enough for
some applications). 
As soon as processes waits for I/O, they are all equal.

I wonder how difficult it was to just add priorities to your I/O scheduler.
Basically, I think everything is there: You sort requests already and we
have deadlines. For sorting scores are used.
So the idea is: Why not just give I/O submitted on behalf of -19 processes
(and RT) some higher score and a shorter deadline than +19 ones?
Probably, it would only affect reads, as there we have the processes wait
on them; writes are often triggered asynchronously anyway.

This should have the effects that we want: When there's no fight for I/O
bandwidth, everybody just gets maximum performance as the I/O scheduler's
queue will be short. As soon as processes fight for reads, unniced processes
have a higher chance of getting served first.

Just think of nightly updatedb on a webserver for a real-world example why
this may matter.

Looks like something not too difficult to do, but my current knowledge on
2.5 code is somewhat sparse :-(

Regards,
-- 
Kurt Garloff  <garloff@suse.de>                          Eindhoven, NL
GPG key: See mail header, key servers                        SuSE Labs
SuSE Linux AG, Nuernberg, DE                            SCSI, Security

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

             reply	other threads:[~2002-10-21  7:20 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2002-10-21  7:26 Kurt Garloff [this message]
2002-10-21  8:24 ` Jens Axboe
2002-10-21  8:51   ` Christoffer Hall-Frederiksen

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=20021021072629.GD6630@nbkurt.casa-etp.nl \
    --to=garloff@suse.de \
    --cc=axboe@suse.de \
    --cc=linux-kernel@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

all inboxes | Powered by JetHome®