Internet Info 1994 March/Internet Info CD-ROM (Walnut Creek) (March 1994).iso/inet/ietf/tuba/tuba-minutes-92nov.txt

Text File | 1993-02-17 | 9KB | 195 lines

Editor's Note: Minutes received after the cut-off for inclusion in the
Proceedings.

Minutes of the TCP/UDP over CLNP-addressed Networks Working Group (TUBA)}
Reported by Peter Ford/LLNL

Keith Sklower (UC Berkeley) reported on his implementation of TUBA
using BSD 4.4 as a software base.  The total number of lines  of added
or changed source code, including ifdefs, turns out to be around 700
lines, with about 200 lines devoted to the TCP/UDP glue layer for
CLNP.   These changes do not include changes to gethostbyname() where
Keith indicated he wanted to get Bill Manning's code.   It was noted
that the API issues, such as gethostbyname() and socket() in Unix, were
very important and that these interfaces were not worked on in the IETF
but in other standards organizations such as IEEE POSIX.

The discussion quickly turned to the issue of what to use for TCP/UDP
connection identifiers.  Ross Callon (DEC) supported the use of the
CLNP NSAP ID field as the TCP connection identifier since  it is
possible to build systems where the ID field does not change when the
NSAP might change due to changes in routing such as the case in some
implementations for mobile hosts.  Keith  explained that using the NSAP
would be better since it places no restrictions on NSAP format.  Brian
Carpenter stated that TUBA has great utility for those who are going to
use TUBA but are not really part of the IP Internet, so there should
not be restrictions placed on NSAP formats for the use of TUBA which
would impede its use in this context.  There were several comments
speculation that perhaps 6 bytes of ID would not be enough in the
future  and that administration of this ID space might prove to be
problematic.    Several people noted that the pseudoheader may have
outlived its usefulness.

Vint Cerf explained the origin and rationale for the TCP pseudoheader.
When it was decided to split the functionality  of old TCP into current
TCP and IP, there  was concern as to how to detect corruption of TCP
connection identifiers.  It would have been possible to directly store the IP
addresses in the TCP header or use the pseudoheader which goes into
the IP header to get IP addresses.  Several people
asked about the probability of this sort of problem, and Vint noted
that something always goes wrong and that TCP has benefited
tremendously by designing accordingly.

The group reached a tentative consensus on an implementation agreement
to use the entire NSAP for the purposes of generating TCP connection
Ids along with the NSEL and TCP/UDP Port numbers.  This will allow trial
implementations to proceed.  Those who want to implement alternatives
are free to do so for experimental testing, but for the sake of
interoperability tests of TUBA, implementors are encouraged
to implement along the lines of the trial implementation agreement.
Keith Sklower is looking into the licensing issues of his
implementation and will report to the TUBA list.

The issue of network translation boxes came up and was quickly tabled as
not being germane to the goals of developing TUBA at this time.
Network translation may  resurface once  TUBA has been completely
specified.

The discussion shifted to the issues TUBA documentation and what TUBA
transition would look like.    Ross  Callon discussed the "son of rfc
1347" which is the core TUBA document.  Ross also noted that his TUBA
addressing internet-draft needs to be updated and noted that some other
topics needed to be discussed in it such as: EIDs, limitation of CLNP
addressing to 20 bytes is not mandatory, the nature of scaling and
flexibility of CLNP NSAPs, and some kind of analysis of future
Internetwork topologies which reflect  possible futures such as telco
provided switching and its impact on the "branchiness" of the network.
Dave Piscitello reported that he is working on updating his document
on the use of CLNP in the TUBA environment.  Bill Manning and Richard
Collela reported they have made significant changes to the  DNS NSAP
resource record RFC and hope to reissue it soon.

A brief discussion of transition issues revealed that they were very
complicated and a second TUBA meeting was scheduled for Friday (20 Nov.
92).

TUBA transition meeting (20 Nov. 92)

The initial topic was the question of whether  to assume that all the
infrastructure would become CLNP capable vrs. the use of  encapsulation
or network layer translation.  The  majority of the group is in favor
of assuming that a ubiquitous CLNP infrastructure can be deployed in
a timely manner and it was noted that this is progressing pretty well.
Brian Carpenter suggested tighter coupling between the TUBA group and
other groups interested in CLNP deployment: HEP, etc.

Bill Manning reported on his NSAP DNS work and reported that  he has an
implementation of BIND  which handles NSAPs.  He has some changes he
still needs to make, but is willing to share his implementation with
others.  He is deploying this for testing on ns.sesquinet.net.

Dave Katz's name was suggested to contribute a document on the use of
ES-IS and dynamic host addressing in the TUBA environment.  Mark agreed
to ask Dave. Frank Solensky mentioned he is studying the use of large
addresses, including SIP, TUBA, PIP, etc., and their impact on APIs and
application interfaces and  hopes to generate a document on these
topics.

Work on new documents was discussed and divided up.  It was pointed out
that the TUBA group needs to do a better job of documenting the
transition process to get to a TUBA world.  Yakov Rekhter has a Mobile
Hosts document he would like to contribute.  Bill Manning, John Curran
and Mark Knopper signed up for a detailed transition plan document.
John Curran, Brian Carpenter and Peter Ford will finish up the response
to the IPv7 requirements fulfillment document pending output from the
requirements BOF held earlier this week.

It was decided a TUBA meeting in late January  would be needed to keep
TUBA progressing.

Attendees

Roland Acra              [e-mail]
Cengiz Alaettinoglu      [e-mail]
Philip Almquist          [e-mail]
Jules Aronson            [e-mail]
Nagaraj Arunkumar        [e-mail]
William Barns            [e-mail]
Bryan Beecher            [e-mail]
Shiraz Bhanji            [e-mail]
David Bolen              [e-mail]
Ross Callon              [e-mail]
Brian Carpenter          [e-mail]
George Chang             [e-mail]
Henry Clark              [e-mail]
Richard Colella          [e-mail]
Michael Collins          [e-mail]
Michael Conn             [e-mail]
John Dale                [e-mail]
Osmund de Souza          [e-mail]
Richard desJardins       [e-mail]
David Dubois             [e-mail]
Jacques Dugast           [e-mail]
Dino Farinacci           [e-mail]
Eric Fleischman          [e-mail]
Peter Ford               [e-mail]
Shoji Fukutomi           [e-mail]
Peter Furniss            [e-mail]
Masayoshi Gohara         [e-mail]
Heather Gray             [e-mail]
Thomas Hacker            [e-mail]
William Haggerty         [e-mail]
Joel Halpern             [e-mail]
Susan Hares              [e-mail]
Tim Howes                [e-mail].
Kathleen Huber           [e-mail]
David Jacobson           [e-mail]
Dwight Jamieson          [e-mail]
Matthew Jonson           [e-mail]
Merike Kaeo              [e-mail]
Akira Kato               [e-mail]
Dave Katz                [e-mail]
Mark Knopper             [e-mail]
John Krawczyk            [e-mail]
Tony Li                  [e-mail]
David Lin                [e-mail]

                                   1
^L
Tracy Mallory            [e-mail]
Bill Manning             [e-mail]
Matt Mathis              [e-mail]
Jun Matsukata            [e-mail]
Donald Merritt           [e-mail]
David Meyer              [e-mail]
Dave Monachello          [e-mail]
Robert Moose             [e-mail]
Dennis Morris            [e-mail]
Jun Murai                [e-mail]
Tu Nguyen                [e-mail]
Andy Nicholson           [e-mail]
Peder Norgaard           [e-mail]
Andrew Partan            [e-mail]
Laura Pate               [e-mail]
David Piscitello         [e-mail]
Yakov Rekhter            [e-mail]
April Richstein          [e-mail]
Henry Sanders            [e-mail]
Dallas Scott             [e-mail]
John Scudder             [e-mail]
Paul Serice              [e-mail]
Keith Sklower            [e-mail]
Frank Solensky           [e-mail]
Richard Thomas           [e-mail]
Paul Traina              [e-mail]
Iain Wacey               [e-mail]
William Warner           [e-mail]
Luanne Waul              [e-mail]
Guy Wells                [e-mail]
Evan Wetstone            [e-mail]
Kirk Williams            [e-mail]
Linda Winkler            [e-mail]
Cathy Wittbrodt          [e-mail]