<NIS.NSF.NET> [IMR] IMR88-11.TXT
NOVEMBER 1988
INTERNET MONTHLY REPORTS
------------------------
The purpose of these reports is to communicate to the Internet Research
Group the accomplishments, milestones reached, or problems discovered by
the participating organizations.
This report is for research use only, and is not for public
distribution.
Each organization is expected to submit a 1/2 page report on the first
business day of the month describing the previous month's activities.
These reports should be submitted via network mail to Ann Westine
(Westine@ISI.EDU) or Karen Roubicek (Roubicek@NNSC.NSF.NET).
BOLT BERANEK AND NEWMAN INC.
----------------------------
REAL-TIME MULTIMEDIA CONFERENCING
We have recently commenced a new project in Research into Real-time
Multimedia Conferencing. The purpose of this project is to
investigate, design and build a prototype infrastructure for
developing multimedia communication applications which use the
DARPA Internet as the communications network. Our development work
will focus on two areas: Multimedia presentation of information and
electronic communication of this information between geographically
separated sites. This topic falls under the more general area of
electronic collaboration technologies.
The creation of such a facility will add several important
dimensions to the modes of communication able to be supported in a
Westine [Page 1]
Internet Monthly Report November 1988
packet switched network. One example of this type application is a
distributed Command and Control situation display which would allow
geographically separated commanders to cooperate in various
decision making activities, using application displays containing
multimedia components. Another example is a real-time multimedia
conferencing facility which would enable users to confer over
computer controlled displays in real-time, together developing
cooperative solutions to problems using a variety of advanced
workstation-based application programs.
To date we have been involved in the following activities:
1 Research into the proper conceptual models for multimedia
communication applications. We have been examining all of the
systems and models known to support real-time collaboration.
In addition, we have examined experienced gained in the DARPA
sponsored Multimedia Conferencing System that was developed by
BBN and ISI.
2. Design and implementation of an improved multimedia
conferencing toolkit which will enable application developers
to build new conferencing applications. We expect to release
a version of this toolkit to the DARPA community in the first
quarter of 1989.
3. Analysis of the issues to be faced by programs that are
intended to be used by groups of users: a group user
interface. To date most conferencing applications have been
single user applications used in a group setting. Our
experience with the DARPA Multimedia Conferencing System has
shown that this is not enough. We are currently exploring the
types of changes to the program operating interface needed in
a conferencing setting.
4. Selection or design and implementation of display oriented
applications to be developed under this infrastructure.
Currently we have designed and partially implemented three new
applications, in addition to the Diamond multimedia document
editor, which can be used as presenters of source material
during a conference:
Maptool: A Map Navigation/Route Planning Tool
VDP: A Video Disk browser tool
Confshell: A multi-user UNIX command shell
Presenter: A talk presentation tool
We are also evaluating a Command and Control system named VIDS
to see if it might be adapted to work in this conferencing
Westine [Page 2]
Internet Monthly Report November 1988
environment.
We are looking for other applications to bring into this
environment either by adaptation or new design.
5. Examination, selection and integration of a variety of new
media types to be used in multimedia communication
applications including digital video and audio media types.
We have designed and are in the process of implementing a
Video Information Server which will allow a user at a
workstation to access video information from a variety of
sources including video disks, tape, cameras and cables.
As these developments result in prototypes we will be
introducing new capabilities into the DARPA Multimedia
Conferencing System. For more information contact Harry
Forsdick (forsdick@diamond.bbn.com).
WIDEBAND NETWORK
The major components of the Wideband Network's monitoring and
control software have been ported from the BBN C/70 computer to the
Sun Workstation environment, allowing monitoring of the network
from Sun Workstations for the first time this month. This
development also supports access to real-time Wideband Network
monitoring data by distributed Cronus operating system components
that use the network for their wide-area communications. The
ported network monitoring code was significantly modified to
simplify its future adaptation to the receipt and processing of new
kinds of monitoring data. Sun-based monitoring and control is
planned during Phase 1 of the Defense Research Internet.
A new multimedia conferencing site was installed at SRI this month.
Including the DARPA, BBN, and ISI sites, there are now four such
sites on the Wideband Network.
The "ESD-Hypercube" ethernet LAN was connected to the Wideband
Butterfly Gateway located at ESD/Mitre facilities. This connection
will support distributed Hypercube simulation activities at ESD and
JPL.
SATNET
No changes have been made on the SATNET during the month of
November. The installation of the new direct link between BBN and
RSRE was delayed because of delays in the installation of the lines
by the PTT's. The lines are now scheduled to be turned over for
testing, by BBN and RSRE on December 2.
Westine [Page 3]
Internet Monthly Report November 1988
INTERNET R&D
We fixed a bug which was causing the gateways at RSRE and UCL to
restart excessively. New software and patches have been sent to
both sites. Also, Bob Hinden attended the DARPA PI meeting in
Dallas/Ft. Worth.
VAX NETWORKING
RFC-1075, describing the multicast routing protocol implemented
this summer was issued.
Bob Hinden (Hinden@BBN.COM)
ISI
---
Internet Concepts Project
Jon Postel visited Dallas-Fort Worth, Texas to attend the DARPA
ISTO PI meeting. Paul Mockapetris attended a WESTNET mtg. in
Boulder, CO., 11 Nov.
Six RFCs were published this month.
RFC 1075: Waitzman, D., C. Partridge, (BBN STC), and
S. Deering, (Stanford), "Distance Vector
Multicast Routing Protocol" November 1988.
RFC 1076: Trewitt, G., and C. Partridge, "HEMS Monitoring
and Control Language", Stanford University,
November 1988.
RFC 1077: Leiner, B., "Critical Issues in High Bandwidth
Networking", Editor, Gigabit Working Group,
November 1988.
RFC 1078: Lottor, M., "TCP Port Service Multiplexer (TCPMUX),
SRI-NIC, November 1988.
RFC 1081: Rose, M., "Post Office Protocol - Version 3", TWG,
November 1988.
Westine [Page 4]
Internet Monthly Report November 1988
RFC 1082: Rose, M., "Post Office Protocol - Version 3, Extended
Offerings", TWG, November 1988.
One article was published:
DeSchon, A., and R. T. Braden, "Background File Transfer
Program (BFTP)", ConneXions -- The Interoperability
Report, Volume 2, No. 12, December, 1988.
Ann Westine (Westine.ISI.EDU)
Los Nettos
All five phase 1 sites are now connected. One redundant path
remains unused because CSU/DSU's are not available. Sites are now
starting to use Los Nettos to get to other member networks. We
have started to EGP with the core advertising accessability to the
Los Nettos backbone (which has no hosts) but none of the member
networks. We will soon advertise primary access to one or more of
our member networks. We have used FTP to exercize the network.
Between two unoptimized Sun workstations, one at ISI and one at USC
we were able to transfer at a rate of 500Kb/s with bandwidth to
spare.
Walter Prue visited UCLA 3 November, to discuss adding a line from
UCLA to SCSC for Los Nettos. Walter Prue made several trips to
UCLA this month to test and debug the Los Nettos connection.
Walter Prue hosted the Los Nettos Technical meeting at ISI, 25
November.
Danny Cohen has developed a proceedure which periodically pings
each interface in the network and can display a map of the network
with non-communicative elements highlighted.
We have had serious support problems with Datatel our CSU/DSU
vendor. There are compatability problems between the Datatels and
our cisco routers. Cisco has been very supportive helping us
resolve problems. Datatel to date has not. However we have
managed to find temporary solutions to our problems.
The favorable T1 tariff change proposed by Pacific Bell appears to
be delayed. More up-to-date information will be available soon.
We can not delay phase 2 waiting for the new tariff.
We have received only one purchase order for phase 2 membership in
Los Nettos. Because of the two to three month lead time on T1
circuits we will not only need a verbal commitment but a Purchase
order soon to get new members operational in time. DARPA has
Westine [Page 5]
Internet Monthly Report November 1988
notified us that most ARPANET connections will be taken down by May
1. The ARPANET connection at ISI will remain operational to serve
Los Nettos members.
Walt Prue (Prue@ISI.EDU)
Multimedia Conferencing Project
This month the multimedia teleconference facility underwent
additional testing for four-site conferences between BBN, DARPA,
ISI, and SRI. These investigations focused on the demands of real-
time video and audio bandwidth and on the throughput of the
Wideband Network.
PVP, the packet video host, was modified and tested to enable it to
interface with the new ST Gateway. This communication will be over
a local net internal to the Butterfly machine.
The latest version of MMCC (the MultiMedia Conferencing Control
program), with autopilot functionality, was installed this month to
facilitate multi-site testing. Redesign of MMCC is underway; use
of the Versatile Message Transaction Protocol (VMTP) is under
consideration as a replacement for MMCC's home-grown multisite
conferencing protocol and collaborative work has begun on the
integration of MMCC and BBN's Diamond/MMCONF system.
Steve attended the DARPA/ISTO Principal Investigators' Meeting in
Dallas/Ft. Worth. Eve and Steve completed a paper about the
multimedia conferencing project for inclusion in the upcoming
special-issue of the ACM SIGOIS Bulletin on Computer-Supported
Cooperative Work.
Dave Walden, Eve Schooler, Steve Casner (djwalden@ISI.EDU,
schooler@ISI.EDU, casner@ISI.EDU)
NSFNET Project
A major task this month was the preparation of a new release of the
NNStat statistics-collection package. Annette DeSchon converted
the NIT interface for the stream-oriented version in SunOS 4.0.
Working with Van Jacobson, we discovered several bugs in Sun's NIT
code. Van Jacobson fixed one bug and suggested a workaround for
others; he reported these bugs and the fixes to Sun. Bob Braden
made a number of minor updates and efficiency improvements in
statspy. We hope to make Release 2.2 of NNStat, including these
changes, available early in December. Later, we will create a
release with other enhancements, including as access control
mechanism. Dave Katz of Merit kindly provided us with the source
Westine [Page 6]
Internet Monthly Report November 1988
for statspy changes for access control; he developed this for
incorporation into the NSFnet backbone NSS's.
Annette also spent time tracking down user problems with BFTP; many
of these proved to be bugs in existing FTP servers. These bugs are
being reported to the responsible (?) parties as they are
discovered. Her work was hampered by very poor Internet
connectivity, especially to East Coast sites. She will bring out a
maintenance release of BFTP soon.
Bob Braden chaired a meeting of the End-to-End Task Force Nov 3-4
at MIT, and worked on updating the Host Requirements RFC in
accordance with agreements at the last IETF meeting.
Bob Braden and Annette DeSchon (Braden@ISI.EDU, DeSchon@ISI.EDU)
MIT-LCS
-------
No report received.
MITRE Corporation
-----------------
No report received.
NTA-RE and NDRE
---------------
No report received.
SRI
---
No report received.
UCL
---
Peter Kirstein attended an EXPRES review meeting, and the meeting
for PIs on DARPA projects.
We have started work in connection with EXPRES and ESPRIT on the
Office Document Architecture, and this is closely coupled with with
our existing X.400 Mail system work, and continued use of the
Diamond multimedia mail system.
John Crowcroft (jon@CS.UCL.AC.UK)
Westine [Page 7]
Internet Monthly Report November 1988
UNIVERSITY OF DELAWARE
----------------------
1. Paul Schragger continues to work on priority and type-of-
service issues in gateways and hosts. Jeff Simpson continues
to work on mechanisms to implement engineered routing
policies, as well as a position paper on existing Internet
administrative policies. Mike Minnich is working on the first
draft of his dissertation. Dave Mills attended the DARPA
Principal Investigator Meeting at DFW airport and the IAB
teleconference at DARPA HQ.
2. We are trying hard to find the cause of frequent routing
spasms that result in our research net 128.4 clocks
disappearing from the NSFNET maps for extended periods up to
two days. The latest evidence implicates an intermittent
routing loop between at least two of the NSFNET backbone NSS
switches, which can't happen in correctly functioning link-
state routing algorithms. Needless to say, we are working this
problem actively.
3. The good folks at Oregon State brought up a Network Time
Protocol server in their backyard and quickly discovered very
bumpy roads to the NSFNET interstate, including fast, long
satellite channels and slow, short 9600-bps country lanes, but
very unstable roadmaps. The result is massive, asymmetric
excursions in delays and delay dispersions that exceed the
tracking range of the phase-lock loop and result in frequent
clock resets. The lesson in this and other recent cases is
that the phase-lock loop must include an adaptive capability
to adjust its tracking parameters and time constants.
4. A paper on multiple-path routing algorithms was incubated,
hatched and will be sent shortly to ACM Computer Communication
Review for fledging. A first-crash award is being claimed as
the first paper submitted to CCR in PostScript format. Another
paper on internetwork time synchronization is going through
some last-minute reference checking.
5. A white paper on high-speed reservation-TDMA networks is going
through final revisions. Our eclectic folks here are now
trying to figure out how to schedule multiple trains on a
richly connected railroad and are discovering this is really
hard. A proposal on engineered routing algorithms is also
being incubated.
Dave Mills (Mills@UDEL.EDU)
Westine [Page 8]
Internet Monthly Report November 1988
NSF NETWORKING
--------------
NSF NETWORKING
UCAR/BOLT BERANEK AND NEWMAN INC., NNSC
Craig Partridge attended the End-to-End Task Force Meeting and
Karen Roubicek participated in the FARnet meeting. Karen gave a
presentation about NSFNET to the Coordinating Committee of the Long
Term Ecological Research (LTER) sites. NSF, NNSC and NCSA are
planning a spring networking workshop for representatives from LTER
sites, which are part of NSF's Division of Biotic Systems and
Resources.
by Karen Roubicek (roubicek@nnsc.nsf.net)
NSFNET BACKBONE (MERIT)
With November over, the NSFNET backbone has been in full production
for five months. Both traffic and network connections have
continued to increase in this period. The number of "allowed"
networks has gone from 173 primary networks and 30 secondary
networks at the beginning of July to 334 primary networks, 148
secondary networks, and 6 tertiary networks by the end of November.
-----------------------------------------------------------
Packets in Packets out
October 356,779,596 443,389,133
November 376,052,359 403,084,221
% change 5.1% -10%
-----------------------------------------------------------
Due to an anomaly at MIDNET in October, the packets out of the
backbone to the mid-level networks reflect a negative percentage
change. On October 15 and 16 MIDNET experienced traffic of 40
million and 25 million packets per day compared to a normal daily
average of less than 1 million. Taking into consideration both the
MIDNET anomaly and low traffic counts reported throughout the week
of the Thanksgiving holiday, the November statistics show a
continuation of the rapid growth in backbone use.
Packets counts are taken at the token ring interface to the E-PSP
in each Nodal Switching Subsystem (NSS) via SGMP. The hourly counts
Westine [Page 9]
Internet Monthly Report November 1988
are collected and stored in a database on the Information Services
host machine.
The D4 framing scheme is presently being used on all 18 trunks of
the NSFNET backbone and test network. We will be converting the D4
to Extended SuperFrame (ESF) in the next two weeks. ESF will
greatly enhance our ability to directly manage the T1 circuits. It
is also a prerequisite to using MCI's Digital Reconfiguration
Services (DRS), to allow dynamic reconfiguration of the backbone
circuit topology.
To improve our ability to monitor traffic flows in the NSFNET
backbone we are about to install a modified version of the NNStat
packets developed at the Information Sciences Institute. In
accordance with our agreement with the National Science Foundation
this tool will help us to gather a net to net traffic matrix. The
monitor will be located on the internal token rings of the NSS.
Modifications necessary to make this work in a NSS environment were
software changes to support the package under 4.3BSD as well as
PROM changes on the token ring boards of the monitoring RT/PC.
by Laura Kelleher (Laura_Kelleher@um.cc.umich.edu)
NSFNET BACKBONE SITES & MID-LEVEL NETWORK SITES
BARRNET
Apple Computer (A/UX development group) and the Monterey Bay
Aquarium Research Institute (MBARI) were added to the net in
November. MBARI has the first known submersible ship-born link to
the Internet; an umbilical connected submarine research vessel
whose mother ship has a direct microwave link to the land-based
MBARINet.
Intermittent problems persist in the (1.528Mb/sec) serial links
between routers and DSU/CSUs, characterized by periodic increases
in receive aborts with occasional paralysis requiring that the CSU
be reset or power-cycled or that some other unnatural acts and
incantations be performed to recover. Variable receive-abort
condition levels accompanied by degraded performance are
experienced on other links as well.
We are in the process of upgrading the CSUs in the BARRNet "spine"
to digital ESF units in the hope that we will eventually be able to
monitor the link transmission quality, at least at the level that
basic ESF provides (which ain't much).
Fun was had by all during the worm-day celebrations, ....repairing
Westine [Page 10]
Internet Monthly Report November 1988
the damage done by one inconsiderate jerk.....yes, Martha, we know
there are holes in Unix.
by Bill Yundt (GD.WHY@forsythe.stanford.edu)
CERFNET
No report received.
CORNELL UNIVERSITY THEORY CENTER
The link between CNUSC (Montpellier, France) and Ithaca is working
but we are not passing packets yet -- CNUSC is still waiting for
parts. We expect to have IP connectivity in late January. The
current plan is for this link to connect directly into the NSFNet
node in Ithaca, although CNUSC will not take part in the IS-IS
routing protocol running on NSFNet, communicating with EGP or its
successor instead. Eventually we hope this link will give the
North American Internet IP connectivity to a number of major
scientific centers in Europe.
Even though we are waiting for new GATED funding to arrive, Jeff
Honig has been spending some time cleaning up and redesigning
existing functions such as the handling of interface up/down events
and timeouts, ICMP redirects, and route table hashing, as well as
memory management. EGP3 will have top priority once the protocol
is clearly defined, and we will be reworking the protocol
interfaces and route handling to make it a more useful tool for
rapid prototyping and experiments in policy-based routing.
Scott Brim has been active in groups handling inter-domain routing
issues, such as the Autonomous Networks Task Force and the IETF's
Interconnectivity Working Group.
by Scott Brim (swb@chumley.tn.cornell.edu)
UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN/NCSANET
No report received.
JOHN VON NEUMANN NATIONAL SUPERCOMPUTER CENTER
No report received.
Westine [Page 11]
Internet Monthly Report November 1988
MERIT/UMNET
During the month of November, Merit changed its ARPANET connection
from HDH to X.25.
by Laura_Kelleher@um.cc.umich.edu
MIDNET
No report received.
MRNET
Internet performance and reachability, as measured by mail
throughput, seemed somewhat worse than in the previous few months.
There was no clear cause, and few complaints.
A few MRNet hosts were infected by the recent worm: the ARPAnet
gateway SUN, several at the University of Minnesota, none
elsewhere. Most members remained connected for the duration. The
organization proved an effective means for exchanging worm-related
information and the ensuing security discussion.
The Executive Committee met in preparation for the December MRNet
General Meeting. This meeting should be an opportunity to describe
MRNet to many potential new members.
The Final Project Report for the $20,000 NSF grant which seeded
MRNet is nearly complete. This document will provide a nice
overview of the purpose, origins, and current status of MRNet.
by Tim Salo, Secretary, MRNet (tjs@uc.msc.umn.edu)
with contributions from Stuart Levy (slevy@uc.msc.umn.edu)
NATIONAL CENTER FOR ATMOSPHERIC RESEARCH AND UNIVERSITY SATELLITE
NETWORK PROJECT
No report received.
NORTHWESTNET
No report received.
NYSERNET
No report received.
Westine [Page 12]
Internet Monthly Report November 1988
OARNET
No report received.
PITTSBURGH SUPERCOMPUTING CENTER
We had no unscheduled network downtime and no significant changes.
by Matt Mathis (mathis@fornax.ece.cmu.edu) SAN DIEGO
SUPERCOMPUTER CENTER
The CERFnet evaluation cisco router has been changed to use IGRP
(from RIP) for the internal routing protocol between SDSC and the
California State University test nodes.
Our direct link (via p4200's) to the University of Hawaii has been
disconnected. They now access the Center via the NSI.
by Paul Love (loveep@sds.sdsc.edu)
SESQUINET
No report received.
SURANET
The following SURAnet sites are presently on-line:
University of Alabama at Birmingham
Alabama Supercomputer Network
University of Alabama at Tuscaloosa
Catholic University of America
Clemson University
Continuous Electron Beam Accelerator Facility
University of Delaware
Department of Energy/Oak Ridge Operations Office
University of Florida
Florida Institute of Technology
Florida State University
Fox Chase Cancer Center
Emory University
Gallaudet University
George Mason University
Georgetown University
George Washington University
Georgia Institute of Technology
University of Georgia
ICASE (Institute for Computer Applications in Science and Engineering)
Westine [Page 13]
Internet Monthly Report November 1988
Johns Hopkins University
University of Kentucky
Louisiana State University
University of Maryland
Mississippi State University
NASA/Goddard
NASA/Langley
National Bureau Of Standards
National Cancer Institute/Frederick Cancer Research Center
National Institute of Environmental Health Sciences
National Institutes of Health
National Radio Astronomy Observatory
National Science Foundation
Naval Research Laboratory
Oak Ridge National Laboratory
Old Dominion University
Supercomputer Research Center (IDA)
University of Tennessee
Triangle Universities Computation Center
Duke University
North Carolina State University
University of North Carolina
Tulane University
Vanderbilt University
Virginia Commonwealth University
US Geological Survey
University of Virginia
Virginia Polytechnic Institute
University of West Virginia
College of William & Mary
SURAnet NETWORKS THAT ARE BEING ADVERTISED TO NSFNET
128.4 DCN
128.8 University of Maryland
128.60 NAVAL RESEARCH LABORATORY
128.61 Georgia Tech
128.82 Old Dominion University
128.109 Triangle Universities
128.140 Emory
128.143 University of Virginia
128.150 National Science Foundation
128.163 University of Kentucky
128.164 George Washington University
128.167 Southeastern University Research Association Network
128.169 University of Tennessee
128.172 Viriginia Commonwealth University
128.173 Virginia Tech
Westine [Page 14]
Internet Monthly Report November 1988
128.175 University of Delaware
128.183 Florida Institute of Technology
128.186 Florida State University
128.192 University of Georgia
128.220 John Hopkins University
128.227 University of Florida
128.231 National Institute of Health
128.239 College of William & Mary
129.2 UMD bogon-net
129.6 National Bureau of Standau30
129.43 National Cancer Institute
129.57 Continuous Electron Beam Accelerator Facility
129.59 Vanderbuilt
129.66 University of Alabama
129.71 West Virginia Net
129.174 George Mason University
129.81 Tulane University
130.11 United States Geological Survey
130.14 National Library of Medicine
130.18 Mississippi State University
130.39 Louisiana State University
130.85 UMBC.NET
130.207 Georgia Institute of Technology
131.118 MINC.NET
192.5.39 University of Delaware
192.5.45 Fox Chase Cancer Center
192.5.57 University of Delaware
192.5.82 Florida State University
192.5.214 DEC
192.5.215 George Mason University
192.5.219 Clemson Univeristy
192.12.121 FSUCS
192.12.122 FSUCS2
192.16.175 Georgetown Univeristy
192.16.176 Louisiana State University
192.26.10 Gallaudet Univeristy
192.26.11 National Research Laboratory-HUBNET1
192.26.12 National Research Laboratory-HUBNET2
192.26.13 National Research Laboratory-HUBNET3
192.26.14 National Research Laboratory-HUBNET4
192.26.17 National Research Laboratory-HUBNET7
192.26.26 National Research Laboratory-FIBER
Westine [Page 15]
Internet Monthly Report November 1988
192.31.192 IDA/Supercomputer Research Center
192.31.193 Catholic University of America
192.33.115 National Radio Astronomy Observatory
192.41.177 SURAnet Network Operations Center
192.42.63 Educom88-Demo
192.42.64 Educom88mail-demo
192.42.142 ICASE-NET
by Jack Hahn (hahn@umdc.umd.edu)
WESTNET
1. We are now running ver. 7 of cisco's operating system, which
includes cisco's official release of SNMP. SNMP is being
tested at the University of Colorado, Boulder (Westnet East),
and the University of Utah (Westnet West).
2. The Datatel (now Dowty) CSU/DSU at NCAR on the 56 kbps serial
line to New Mexico Technet (NMT) was replaced this month.
This link had always been subject to high error rates,
resulting in reduced performance and, at times, users even
being precipitously logged-off by the host computer. The
replacement of the CSU/DSU has apparently fixed the problem.
3. We are running our first X.25 circuit between a cisco gateway
at Colorado State University and a CCI gateway at the
University of Northern Colorado. Bringing the circuit up was
interesting as many of the parameters in the configuration
file required "tweaking," but the connection seems to be
working well now.
4. Three new networks have been added during the month, all in
New Mexico. These include Apache Point Observatory and the
National Sunspot Observatory (both connected through New
Mexico State University), and the Sante Fe Institute
(connected through New Mexico Technet).
5. The annual Technical Group meeting was held Nov. 9-11 at the
University of Colorado at Boulder, and was coordinated by
Carol Ward who did an outstanding job. The meeting was well
attended and informative. Notable among the accomplishments
was the working session on the division of NIC/NOC activities.
Those interested may obtain a preliminary draft of the
"strawman" report from Carol (cward@spot.colorado.edu).
by Pat Burns (pburns@super.org)
Westine [Page 16]
Internet Monthly Report November 1988
TASK FORCE REPORTS
------------------
APPLICATIONS -- USER INTERFACE
The following constitutes a more complete summary of the
meeting of the UITF Video Working Group held last May at the
MIT Media Lab. (A shorter version was included in the July
Internet Monthly). This summary was prepared by Steve Casner
of ISI.
The focus of the meeting was on "video in workstations" --
what's the state of the art, and how will it be used.
Presentations ranged from video window hardware and software
to packet transmission of video to video tools and
applications. Martin Levy of Parallax, Dave Stewart of Sun,
David Milway of Olivetti all described their hardware for
video windows on workstations. Video window functions have
been added to X-Windows and NeWS, but development is still
underway. Glen Reitmeier of the David Sarnoff Research Center
described the DVI (Digital Video Interactive) system; though
the initial target is CD-ROM-based applications, they view DVI
as an architecture that could be applied to workstation video
windows as well.
Milway's group is also working on packet transmission of video
on the Cambridge Fast Ring, and Dan Wilson of Bellcore is
transmitting packet video on Ethernet, both at megabit rates
with only simple compression methods for now. Steve Casner
reported on the multimedia conferencing system using the
WBnet, now supporting simultaneous display of video from
multiple sites in quadrants of the screen; each video channel
is compressed to 128Kb/s. Martin Vetterli of Columbia
University described the requirements for video coding for
packet transmission, and proposed a "sub-band coding"
technique to produce a variable data rate with layers of
prioritized data.
Harry Forsdick of BBN Labs said he would like to integrate
video windows into workstation-to-workstation conferencing.
Dick Phillips, appearing on tape, described his "scientific
visualization" system at Los Alamos using Parallax and pNeWS
to display and capture Cray-generated simulation images on the
workstation. Walter Bender and Glorianna Davenport of MIT
Media Lab talked about structured video (e.g., documentaries).
Ben Davis and Matt Hodges described the role of workstation
video for "image learning" in MIT Project Athena.
Westine [Page 17]
Internet Monthly Report November 1988
The afternoon discussion concentrated on video applications:
How do we use it? It's too early to define the taxonomy of
uses for video (and advantages over other media).
Participants in the meeting are working individually on a
range of video problems, but no group has yet brought all
these pieces together to form, for example, a multimedia
workstation conferencing system. We need to try these
applications to measure how useful they are. For video to be
used as other media are now, there is a strong need for
authoring tools. People have grown accustomed to high-quality
production in video, making the task even more difficult. The
need is equally strong for reading tools; structured
multimedia documents will have dual control, both by the
author defining how the pieces form the whole and by the
reader selecting the parts of interest (e.g., text to explain
an event in the video or a video segment to illustrate an
example in text)".
In discussing the above points subsequent to the VWG meeting,
the full UITF developed the following list of potential
application areas: education, command and control, military
applications, process control (monitoring), and medical
applications. There was also considerable sentiment that,
because of our high expectations (with respect to production
quality), we would not see much archival use of the sort of
"casual video" that comes from running around with a HandyCam
or that resulting from a real-time teleconference. Finally,
several members voiced the need to address compression vs.
quality as a continuum, rather than assuming some fixed
bandwidth and trying to squeeze the quality down, or assuming
some fixed quality (e.g. broadcast tv) and trying to squeeze
the bandwidth down.
Keith Lantz (LANTZ@ORC.OLIVETTI.COM)
AUTONOMOUS NETWORKS
Several members of the ANTF attended Barry Leiner's second
"interconnectivity security" workshop at MIT in November. It
was an interesting and productive meeting. In particular, we
generated a list of short and long term projects for ANTF and
others to pursue in the general area of "policy routing for
resource sharing across interconnected autonomous networks". A
report on the meeting is being prepared by Leiner. I have
prepared a very informal note for the ANTF on the more open-
ended issues that arose at the meeting and which deserve
further discussion in the task force.
Westine [Page 18]
Internet Monthly Report November 1988
We are continuing to refine and elaborate our articulation of
these issues and will meet again in February at ISI.
Deborah Estrin (Estrin@OBERON.USC.EDU)
END-TO-END SERVICES
The task force met at MIT Laboratory for Computer Science on
Nov 3-4, 1988. The meeting was enlivened by the concurrent
de-worming that Dave Clark's graduate students were performing
on the MIT networks.
IP MULTICASTING
Dave Waitzman of BBN completed an initial version of a RIP-
based multicast router and submitted an RFC on it (RFC-1075),
before funding was exhausted. This code will be distributed
by Stanford. BBN is said to be close to completing an SPF-
based multicast routing algorithm in the Butterfly gateways.
Both the distance-vector and link-state algorithms were
described in general terms by Steve Deering in his SIGCOMM '88
paper. Steve's conclusion is that link-state algorithms are
more suitable for multicast routing than are distance-vector
algorithms.
VMTP
Dave Cheriton put out a new release of VMTP for Unix (4.3BSD
and SunOS 4.0) This release includes major cleanup of VMTP
code and many simplifications owing to its use of recursion
(as described in Dave's SIGCOMM '88 paper). This version
lacks only security and streaming.
The task force discussed the procedure for moving VMTP to
draft standard status.
FAST HOST INTERFACES
A number of End-to-End task force members are interested in
the the architecture of host interfaces, especially for high-
speed networking. Dave Clark has written a paper on this
subject, and Eric Cooper of CMU is actually building a high-
performance host interface board using very similar ideas.
Dave Cheriton is working on his NAB, a high-speed interface
board that uses its knowledge of VMTP formats to perform
pipelined checksumming and encryption. The task force agreed
to make host interfaces an official item of business.
Westine [Page 19]
Internet Monthly Report November 1988
PERFORMANCE
Dave Clark reported on a paper he wrote with Romkey and
Salwen, studying the theoretical performance limit of an
actual TCP implementation. Their conclusion agreed with Van
Jacobson's, that the ultimate limit on TCP performance will be
set by memory bus time to move and checksum the data, not by
the protocol processing.
There are a number of studies under way on congestion control
and rate-based flow control algorithms in the Internet: Van
Jacobson, Dave Clark, and Dave Cheriton all have students
working in this area.
The task force discussed gateway performance issues,
distinguishing "soft state" from "caching". Soft state is
visible to the higher layers but can be dynamically recreated
after a crash. A cache speeds gateway processing while being
totally transparent to the architecture. Clark's measurements
show that in current IP gateways the IP-specific processing is
actually negligible (10%), so caching is not needed; however,
it will become vital to achieve high gateway throughputs as
more complex gateway algorithms are introduced for accounting,
policy-based routing, and TOS routing and queueing. There is
some evidence that for stable and fair capacity allocation
under rate-based end-to-end flow control, soft state will be
needed in the gateways.
TIME TO LIVE
The task force discussed the ambiguous use of the IP TTL field
as both a hop count at the IP level and to bound TCP segment
lifetimes. There was a feeling that a transport protocol
ought to provide for bounding lifetimes itself; this implies
the necessity of universal synchronized clocks. It was
suggested that a future transport protocol design should start
from the assumption of synchronized clocks, and might be
simplified as a result.
Bob Braden (Braden@ISI.EDU)
Westine [Page 20]
Internet Monthly Report November 1988
INTERNET ARCHITECTURE
The INARC Task Force meeting scheduled for 10-11 January 1989
has been cancelled. An Internet research workshop is to be
held later in the year sponsored by the ACM SIGCOM and INARC
Task Force, which will provide both wide exposure and strong
constituency. As previously planned, the workshop will include
research presentations, solicited and unsolicited, from
throughout the Internet community. Important areas of
interest include policy-based technologies, advanced routing
architectures, high-speed networks and interfaces, network
management, congestion avoidance/control and advanced
transport protocols. An announcement is to appear in ACM
Computer Communication Review.
Dave Mills (Mills@HUEY.UDEL.EDU)
INTERNET ENGINEERING
The next meeting of the IETF has been scheduled for January
18-20, 1988 at the University of Texas. Additional details
and a preliminary agenda will be distributed to the IETF
mailing list. Specific requests for information can sent to
bowers@sccgate.scc.com. Requests to be added to the IETF-
interest mailing can be sent to ietf-request@isi.edu.
Phill Gross (gross@sccgate.scc.com)
INTERNET MANAGEMENT
No report received.
PRIVACY
During November, John Linn and Steve Kent worked
(respectively) on revisions to RFC-1040 and to its companion
certificate-based key management RFC. Drafts were distributed
to the privacy task force membership before the Thanksgiving
holiday and will be reviewed at a one-day task force interim
meeting to be held at BBNCC on 6 December. We anticipate
broader release of this pair of RFCs early in 1989.
John Linn (Linn@CCY.BBN.COM)
Westine [Page 21]
Internet Monthly Report November 1988
ROBUSTNESS AND SURVIVABILITY
No report received.
SCIENTIFIC COMPUTING
No report received.
DSAB
----
No progress to report this month.
Charlotte Tubis (Tubis@Purdue.Edu)