RE: XXX-implementors?

"Douglas Otis" <dotis@sanlight.net> Mon, 15 April 2002 22:05 UTC

X-Sent: 15 Apr 2002 22:05:18 GMT
From: "Douglas Otis" <dotis@sanlight.net>
To: <tcp-impl@grc.nasa.gov>
Subject: RE: XXX-implementors?
Date: Mon, 15 Apr 2002 15:05:13 -0700
Message-ID: <NEBBJGDMMLHHCIKHGBEJCEHBDBAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0)
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
In-Reply-To: <200204152017.UAA07600@gra.isi.edu>
Importance: Normal
Sender: owner-tcp-impl@grc.nasa.gov
Precedence: bulk
Status: RO
Content-Length: 1521
Lines: 39

Bob,

The tsvwg reflector (tsvwg@ietf.org) actively provides feedback on SCTP as
there are several outstanding related documents still in process.  There
appears to be a desire to introduce many of the framing like features of
SCTP into TCP by those wishing to incorporate in high level NICs either
iSCSI or iWARP applications.  This may suggest a common reflector would be
desired.  SCTP and TCP should ensure network fairness if algorithms change.

These TCP (SCTP like) features are:

 - Aligned placement headers with isolated verification/placement/integrity
checks used to handle internal routing of data payloads checked on the fly.

 - Implied TCP framing verification.  (Would a dual CRC (header/payload)
provide needed error checks against framing loss?)

 - TLV bundling of objects for maximal packet size, or will small packets at
> 1Gb/s data rates not be a concern?

If modifications of TCP move forward, then the difference between TCP and
SCTP grows small and yet differences in implementation may provide
justification such as:

 - Should TCP be optimized for Ethernet or InfiniBand for future data rates.

Doug

On April 15, 2002 1:17 PM Bob Braden (braden@ISI.EDU) wrote:
>
> I have an organizational question.  This mailing list concerns the
> implementation of TCP.  Is there another mailing list for implementation
> issues of SCTP?  I can imagine that there might be some overlap, e.g.,
> in congestion control issues; should this mailing list be tsv-impl
> instead of tcp-impl??
>
> Bob Braden
>

XXX-implementors? Bob Braden
RE: XXX-implementors? Douglas Otis