All of lore.kernel.org
 help / color / mirror / Atom feed
From: Eddie Kohler <kohler@cs.ucla.edu>
To: dccp@vger.kernel.org
Subject: Re: [RFC] dccp ccid-3: High-res or low-res timers?
Date: Mon, 17 Nov 2008 19:27:29 +0000	[thread overview]
Message-ID: <4921C5A1.5030609@cs.ucla.edu> (raw)
In-Reply-To: <20081115105042.GA7798@gerrit.erg.abdn.ac.uk>

Gerrit Renker wrote:
> I would appreciate some advice and insights regarding the use of
> high-resolution timers within a transport protocol, specifically
> DCCP with CCID-3 (RFC 5348).
> 
 > ...
 >
> Summing up, I have doubts that basing CCID-3 will bring advantages and
> would much rather go the other way and (consistently) use lower resolution.
> 
> Thoughts?

I agree.  If one way must be chosen, then choose lower resolution timers.

The biggest potential problem with lower-resolution timers is that a sender's 
rate might be limited, not by network characteristics, but by timer 
resolution.  But DCCP allows a fair amount of burstiness already.  And there 
may be ways to avoid rate limitation in common cases without resorting to 
hrtimers.  For example, a sending application could use a mixture of 
non-blocking system calls, allowing the sending application to "poke" the DCCP 
implementation on every scheduling.

At any rate, it seems worth trying.

Eddie


WARNING: multiple messages have this Message-ID (diff)
From: Eddie Kohler <kohler@cs.ucla.edu>
To: Gerrit Renker <gerrit@erg.abdn.ac.uk>,
	Leandro Sales <leandroal@gmail.com>,
	Arnaldo Carvalho de Melo <acme@redhat.com>,
	ian.mcdonald@jandi.co.nz, DCCP Mailing List <dccp@vger.kernel.o
Subject: Re: [RFC] dccp ccid-3: High-res or low-res timers?
Date: Mon, 17 Nov 2008 11:27:29 -0800	[thread overview]
Message-ID: <4921C5A1.5030609@cs.ucla.edu> (raw)
In-Reply-To: <20081115105042.GA7798@gerrit.erg.abdn.ac.uk>

Gerrit Renker wrote:
> I would appreciate some advice and insights regarding the use of
> high-resolution timers within a transport protocol, specifically
> DCCP with CCID-3 (RFC 5348).
> 
 > ...
 >
> Summing up, I have doubts that basing CCID-3 will bring advantages and
> would much rather go the other way and (consistently) use lower resolution.
> 
> Thoughts?

I agree.  If one way must be chosen, then choose lower resolution timers.

The biggest potential problem with lower-resolution timers is that a sender's 
rate might be limited, not by network characteristics, but by timer 
resolution.  But DCCP allows a fair amount of burstiness already.  And there 
may be ways to avoid rate limitation in common cases without resorting to 
hrtimers.  For example, a sending application could use a mixture of 
non-blocking system calls, allowing the sending application to "poke" the DCCP 
implementation on every scheduling.

At any rate, it seems worth trying.

Eddie


  parent reply	other threads:[~2008-11-17 19:27 UTC|newest]

Thread overview: 59+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-09-29 18:33 Protocol not attached Leandro Sales
2008-09-29 18:37 ` Arnaldo Carvalho de Melo
2008-09-30 18:32 ` Leandro Sales
2008-09-30 19:16 ` Arnaldo Carvalho de Melo
2008-09-30 21:24 ` Leandro Sales
2008-09-30 22:11 ` Arnaldo Carvalho de Melo
2008-10-01 18:27 ` Leandro Sales
2008-10-01 19:18 ` Leandro Sales
2008-10-01 19:37 ` Arnaldo Carvalho de Melo
2008-10-01 19:43 ` Leandro Sales
2008-10-01 21:38 ` Leandro Sales
2008-10-01 22:20 ` Ian McDonald
2008-10-02  4:47 ` Gerrit Renker
2008-10-03  5:44 ` Leandro Sales
2008-10-03 13:24 ` Arnaldo Carvalho de Melo
2008-10-03 14:18 ` Arnaldo Carvalho de Melo
2008-10-11  7:35 ` Gerrit Renker
2008-10-11 17:28 ` Leandro Sales
2008-10-13 14:51 ` Gerrit Renker
2008-10-16 21:09 ` Leandro Sales
2008-10-17  6:06 ` Gerrit Renker
2008-10-18 11:06 ` Gerrit Renker
2008-10-20 13:13 ` Arnaldo Carvalho de Melo
2008-10-21 14:19 ` Leandro Sales
2008-10-21 14:32 ` Arnaldo Carvalho de Melo
2008-10-21 18:08 ` Leandro Sales
2008-10-22  1:37 ` Leandro Sales
2008-10-22  1:37 ` Leandro Sales
2008-10-22 11:38 ` Arnaldo Carvalho de Melo
2008-10-22 13:30 ` Leandro Sales
2008-10-22 15:55 ` Gerrit Renker
2008-10-22 19:27 ` Leandro Sales
2008-10-23  5:18 ` gerrit
2008-11-06 13:38 ` Leandro Sales
2008-11-06 15:20 ` Gerrit Renker
2008-11-06 15:38 ` Arnaldo Carvalho de Melo
2008-11-06 17:46 ` Leandro Sales
2008-11-06 18:04 ` Leandro Sales
2008-11-06 18:17 ` Leandro Sales
2008-11-06 22:05 ` Leandro Sales
2008-11-08  8:50 ` Gerrit Renker
2008-11-15 10:50   ` [RFC] dccp ccid-3: High-res or low-res timers? Gerrit Renker
2008-11-15 10:50     ` Gerrit Renker
2008-11-16  8:14     ` Ian McDonald
2008-11-16  8:14     ` Ian McDonald
2008-11-17  6:48     ` David Miller
2008-11-17  6:48       ` David Miller
2008-11-17 19:27     ` Eddie Kohler [this message]
2008-11-17 19:27       ` Eddie Kohler
2008-11-17 21:16     ` [dccp] [RFC] dccp ccid-3: High-res or low-res timers? <cross Gorry Fairhurst
2008-11-17 21:16       ` [RFC] dccp ccid-3: High-res or low-res timers? <cross post> Gorry Fairhurst
2008-11-18  5:07     ` [RFC] dccp ccid-3: High-res or low-res timers? Gerrit Renker
2008-11-18  5:07       ` Gerrit Renker
2008-11-18  6:14     ` Gerrit Renker
2008-11-18  6:14       ` Gerrit Renker
2008-11-18 17:41     ` Ian McDonald
2008-11-18 17:41       ` Ian McDonald
2008-11-20  6:24     ` Gerrit Renker
2008-11-20  6:24       ` Gerrit Renker

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=4921C5A1.5030609@cs.ucla.edu \
    --to=kohler@cs.ucla.edu \
    --cc=dccp@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 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.