Re: [Tsvwg] viewgraphs for a talk in tsvwg in Yokohama tomorrow

Marco Mellia <mellia@prezzemolo.polito.it> Thu, 18 July 2002 13:25 UTC

Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged)) by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA02016 for <tsvwg-archive@odin.ietf.org>; Thu, 18 Jul 2002 09:25:15 -0400 (EDT)
Received: (from daemon@localhost) by optimus.ietf.org (8.9.1a/8.9.1) id JAA05194 for tsvwg-archive@odin.ietf.org; Thu, 18 Jul 2002 09:26:14 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1]) by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA02697; Thu, 18 Jul 2002 08:58:57 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176]) by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA02666 for <tsvwg@optimus.ietf.org>; Thu, 18 Jul 2002 08:58:55 -0400 (EDT)
Received: from tstat.tlc.polito.it (IDENT:0/7D1OQ3nIBXIdqTvgMiPyclMlIyBcFX@verza.polito.it [130.192.9.150]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA01067 for <tsvwg@ietf.org>; Thu, 18 Jul 2002 08:57:55 -0400 (EDT)
Received: from localhost (localhost [[UNIX: localhost]]) by tstat.tlc.polito.it (8.11.6/8.11.6) id g6ICwfn05262; Thu, 18 Jul 2002 14:58:41 +0200
Content-Type: text/plain; charset="iso-8859-1"
From: Marco Mellia <mellia@prezzemolo.polito.it>
To: Sally Floyd <floyd@icir.org>, tsvwg@ietf.org
Subject: Re: [Tsvwg] viewgraphs for a talk in tsvwg in Yokohama tomorrow
Date: Thu, 18 Jul 2002 14:58:40 +0200
User-Agent: KMail/1.4.1
References: <200207171141.g6HBfeS77718@cougar.icir.org>
In-Reply-To: <200207171141.g6HBfeS77718@cougar.icir.org>
Cc: tsvwg@ietf.org
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200207181458.40937.mellia@mail.tlc.polito.it>
Content-Transfer-Encoding: 8bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

Sally,
may be I'm a bit confused, and I hope that you can help me.

My concern about HighSpeedTCP is related to the value of the cwnd you take as 
example.
Nowadys, it seems that the maximum wnd used during transmission by real 
systems is about 32* or 64* MSS (don't forget the rwnd that limits the wnd 
value...). And this is because the end-systems must keep track of all the 
transmitted segments that fit into current transmission window, so that in 
case of packet loss they can retransmit the lost segment.

Neglecting all the possible bottlenecks, but just considering the amount of 
memeory required to keep track of the in fligh packets, we have

 83333*1500 bytes ~ 119Megabytes!

On both sender and receiver side... for every TCP connection
Is this realistic?

I personally don't feel as a big problem the fact that it takes too much time 
to reach window as large as 83333. Simply no end system will ever be able to 
exploit 10Gbps speed with RTT 100ms.

Unless we radically change the sliding window protocol that is at the TCP 
root.

Hope this helps.

+-----------------------------------+
| Marco Mellia - Assistant Professor|
| Tel: 39-011-5644173               |
| Politecnico di Torino             |
| Corso Duca degli Abruzzi 24       |
| Torino - 10129 - Italy            |
| http://www1.tlc.polito.it/mellia  |
+-----------------------------------+

> I am giving a talk in tsvwg tomorrow in Yokohama on the two
> drafts on HighSpeed TCP:
>
> HighSpeed TCP for Large Congestion Windows,
> draft-floyd-tcp-highspeed-00.txt
>
> and:
>
> Limited Slow-Start for TCP with Large Congestion Windows,
> draft-floyd-tcp-slowstart-00.txt
>
> The viewgraphs are at:
> http://www.icir.org/floyd/talks/floyd-tsvwg-Jul02.ps
> http://www.icir.org/floyd/talks/floyd-tsvwg-Jul02.pdf
>
> There is also a HighSpeed Web Page at:
> http://www.icir.org/floyd/hstcp.html
>
> ------
>
> If there is time, I might also talk briefly about a somewhat
> more speculative draft:
>
> Quick-Start for TCP and IP,
> draft-amit-quick-start-00.txt
>
> The viewgraphs for this are at the end of the other viewgraph file.
>
> Quick-Start is designed to allow connections to start up more
> quickly than slow-start, after a pass of feedback from all of
> the routers along the path.
>
>
> Thanks,
> - Sally

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg