Hi,
I support the suggestion of expanding the document to include link
layer characteristics so
transport layer designers understand the link layer issues.
It would also be hard to imagine that many link layers will adapt to
TCP requirements. It would
seem more of a managable problem to have TCP adapt to different link
behaviors. Therefore, the
suggestion below to have the document include the link layer
characteristics of various communication
links is a very good one.
Emad.
> ----- Begin Included Message -----
>
> From owner-pilc@lerc.nasa.gov Mon Dec 14 17:37:53 1998
> X-Authentication-Warning: assateague-fi.lerc.nasa.gov: listserv set sender
> to owner-pilc@lerc.nasa.gov using -f
> X-Info: IDE / NASA Lewis Research Center
> X-Sender: caivanc@popserve.lerc.nasa.gov
> Date: Mon, 14 Dec 1998 11:26:09 -0500
> To: pilc@lerc.nasa.gov
> From: William D Ivancic <William.D.Ivancic@lerc.nasa.gov>
> Subject: TCP over wireless type links
> Mime-Version: 1.0
> Content-Transfer-Encoding: quoted-printable
>
> Relative to discussion at the PILC BOF, I strongly support generation of a
> document that indicates the basic link requirements that current TCP
> implementations expect so that link designers can properly engineer the
> links (when possible). It would also be useful to document general
> characteristics of various communication links - particularly satellite
> and
> wireless links so that transport protocol implementers understand some of
> the problems that the link layer designers have. Information that
> addresses the difficulties in meeting all current TCP link layer
> requirements would be useful.
>
> During the IETF performance implications of link characteristics BOF, some
> insightful comments were made regarding TCP's control algorithms and the
> changing characteristics of the Internet. Many unique networks were
> presented some with various common elements. To paraphrase the
> statement,
> control systems can be well design for a bounded problem with bounded
> parameters. When those parameters are exceeded, the system may become
> unstable. Current TCP works fairly well for much of today's Internet, but
> does exhibit shortcomings in certain situations such as low-data-rate
> low-RF-bandwidth wireless communications.
>
> For moderate and high-data-rate links (greater than 10s of kbps),
> techniques to provide data links that meet the requirement of current TCP
> are fairly well understood and reasonable doable (i.e. concatenated
> convolutional and Reed-Solomon coding, interleaving, correction for
> erasures, etc...) In additions, some minor modifications may help improve
> overall performance. However, these techniques cannot be practically
> applied to low-data-rate low-RF-bandwidth (10s of kbps and less).
> Furthermore, some of these links experience short outages and fades.
> Thus,
> it is highly probable that such links will require a different transport
> protocol or major modifications to current TCP in order to perform
> efficiently (both in usable data throughput and RF bandwidth utilization).
> Therefore, I suggest that we consider separating the low-data-rate
> low-RF-bandwidth wireless problems for those of moderate and
> high-data-rate
> communication links (wireless or otherwise).
>
>
> *********************************************
> William D. Ivancic
> NASA Lewis Research Center
> 21000 Brookpark Rd. MS 54-8
> Cleveland, Ohio 44135
> USA
> Phone: 1 216 433 3494
> FAX: 1 216 433 8705
> Email: William.D.Ivancic@lerc.nasa.gov
> wivancic@lerc.nasa.gov
> *********************************************
>
> ----- End Included Message -----
This archive was generated by hypermail 2b29 : Mon Jan 28 2002 - 09:12:19 EST