<NIS.NSF.NET> [IMR] IMR87-05.TXT
Westine [Page 1]
~
MAY 1987
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@SH.CS.NET).
BBN LABORATORIES AND BBN COMMUNICATIONS CORPORATION
---------------------------------------------------
ARPANET/MILNET
RFC 1005 was released in May. It describes the proposed AHIP-E
enhancement to the AHIP (1822) protocol. This will affect every
AHIP and HDH host on the MILNET and the ARPANET. Comments are
solicited, to the mailbox "ahipe@bbn.com".
Westine [Page 1]
Internet Monthly Report May 1987
DIAMOND MULTIMEDIA COMMUNICATIONS
BBN and ISI have started a weekly meeting discussing Multimedia
Communications. We are making use of the Wideband Net Video, Voice
and MMConf multimedia conferencing facilities to both handle the
communication requirements of this distributed meeting as well as
to test and evaluate these systems. All of the participants in the
meetings have access to Diamond for preparing material to be
presented prior to the meetings. We use the Diamond X.400 mail
facilities to exchange copies of discussion material in advance.
During the conference, Diamond multimedia documents can be
displayed and simultaneously edited at both meeting sites. The
Wideband Network Packet Video system provides a full motion black
and white video image of the other site which is displayed on a
television monitor at each site. We are also using a full duplex
Packet Voice system for vocal communications. Meeting participants
wear lightweight headsets connected to small FM radios that receive
the audio from the other site.
This facility has been used several times in the past month for a
variety of purposes including a meeting of the IAB, an all-day
seminar originating at ISI on the occasion of their 15th
anniversary, technical discussions over an evolving protocol
document, as well as the weekly Multimedia Communications meetings
between ISI and BBN.
WIDEBAND NETWORK
The performance of the Wideband Network sites' satellite channel
subsystems has been stable since the network was moved to CONTEL
ASC's space segment at the end of April. The move of Wideband
earth station monitoring and control to CONTEL ASC's facility in
Atlanta, GA was completed during May.
A coast-to-coast Internet Activities Board meeting was supported by
the Wideband Network and the multimedia conferencing facilities at
BBN and ISI on May 7. The system worked flawlessly throughout the
entire day-long meeting.
Enough BSMI boards have been successfully tested for to complete
BSMI deployment to all of the BSATs in the field. BSMI
installations were done at ISI, SRI, M/A-COM, and CMU during the
month. The final BSMI installations (DCEC, RADC, and Ft. Monmouth)
are scheduled for June.
Westine [Page 2]
Internet Monthly Report May 1987
SATNET
The SATNET continued to perform well. There was only one SATNET
outage and three minor gateway/line outages. Task force work is
continuing on measurements of SATNET performance.
GATEWAYS
The Gateways continue to be stable. The Internet is now up to
about 215 active networks. We are now debugging our
implementations of IP reassembly for the Butterfly and LSI-11
gateways. We expect to begun fielding this in the near future.
We have begun working on implementing the SURAP protocol to
interface the Butterfly Gateway to the SURAN network. We have two
LPR's at BBN and plan to set them up as a small test network to
test out the gateway SURAP implementation.
Bob Hinden
ISI
---
Internet Concepts Project
Ray McFarland from NSA visited ISI on May 6, to give a seminar
on advanced research topics in computer security including:
expert systems and artificial intelligence, massively parallel
systems architectures, tightly coupled distributed systems,
and protocl engineering diciplines. Jon Postel attended
Network Meetings at NRI, in Washington D. C. May 12-15.
Four RFCs were published:
RFC 1005: Khanna, A., and A. Mails, "The ARPANET-AHIP-E Host
Access Protocol (Enhanced AHIP)".
RFC 1006: Rose, M., and D. Cass, "ISO Transport Service on
Top of the TCP Version: 3".
RFC 1010: Reynolds, J., and J. Postel, "Assigned Numbers".
RFC 1011: Reynolds, J., and J. Postel "Official Internet
Protocols".
Ann Westine (Westine@ISI.EDU)
Westine [Page 3]
Internet Monthly Report May 1987
Multimedia Conferencing Project
The third teleconference meeting of the IAB was held May 7,
with six members at ISI and seven at BBN. All the subsystems
involved in the voice and video communication demonstrated a
significant improvement in reliability as everything worked
smoothly for nine hours of continuous operation! In
particular this was due to the delivery of planned
enhancements in the Wideband Network hardware and software.
This meeting was the inaugural use of the new teleconference
room at ISI. We hope the availability of this dedicated room
will provide more opportunities for others to make use of our
teleconference system.
Steve Casner (Casner@ISI.EDU)
Brian Hung's new program for the IBM-PC AT can now send both
bitmap and text media in the Diamond document format. The
program is being tested, and Brian has successfully sent
multimedia messages containing bitmap and text to Steve Casner
and Joyce Reynolds. Currently the text editing facility is
very limited. Brian plans to look at ways of adding some text
editing capability and also making some changes to the current
program to make it more easy to use.
Joyce Reynolds has begun active exchanges of Diamond 3.0
Multimedia mail with BBN.
Brian Hung and Joyce Reynolds (Hung@ISI.EDU, Reynolds@ISI.EDU)
NSFNET Project
Bob Braden, with Milo Medin of NASA, visited Vitalink in order
to understand in depth the Translan technology, and in
particular its routing and O&M facilities. The following week
they presented a report on this to a meeting of the NPAG-TC
group at NSF, concerned with the future technology for the
NSFNET backbone. Bob Braden also attended the first meeting of
the Internet network management working group which is forming
under Dan Lynch's flag, participated in a one-day IAB
teleconferencing meeting, and attended a three-day meeting of
the IRI-Engineering task force held at NRI in Reston, VA.
Under the NSFNET contract, work continued on finishing RFC985
and on gathering and mapping topology data on NSFNET. We
believe that we finally have a map which is correct, although
it is already incomplete.
Bob Braden
Westine [Page 4]
Internet Monthly Report May 1987
Supercomputer and Workstation Communication Project
Alan Katz spent some time investigating NeWS, Sun's
Network/exensible Windowing System and Postscript, which is
the imaging language upon which it is based. Alan gave an
Internet Seminar on NeWS, Postscript, and X-Windows.
Alan helped to evaluate Interleaf's new UPS system, especially
their equations package. Under this system, one can enter
equations in EQN-like format, but can then edit them WYSIWYG
style. Alan Katz attended a conference at UCLA on high
temperature super conductors May 8.
Alan Katz (Katz@ISI.EDU)
MIT-LCS
-------
We are considering an effort to build an interactive, animated
network simulator. We certainly feel the need for such a handy
tool ourselves; we also expect it to be useful to the internet
community (as Van Jacobson has suggested, everyone working on
network protocols seriously should have such a tool at hand).
The goal is to make a simple, flexible/extendible simulator. To
prompt portability, it will be coded in C, running under Berkeley
UNIX 4.x, and using the X window system as the graphic user
interface.
Here is our current (and very preliminary) thought on the simulator
model: each type of network components --network/link, switch,
host, connection (sorry, a bad word), and packet-- will be coded as
separate modules; each module has a number of parameters that can
be specified (e.g. the link has parameters of delay, throughput,
transmission error rate, and type [whether it is a point-to-point
link or a multi-drop channel]; connections have parameters such as
type, data generation pattern and rate). Hooking up a number of
switches and hosts by links (using the mouse to draw them on a
graphic screen), you can build a network; hooking some connections
to hosts, you make data sources and sinks, and then packets can
start flow. Each module is replaceable by another one with the
same interface.
The simulator will make no assumption about either the network or
the traffic models; it will try to provide a library of commonly
seen models. For example, to model network traffic, we may provide
Poisson arrival, bursty or constant rate, etc. models as packet
generators, and bimodal or constant or exponential models as packet
length distributions. The decision on what model to use and the
model validity are the simulator user's consideration. He can code
Westine [Page 5]
Internet Monthly Report May 1987
his own data generators if none in the library fits.
We would like to get a consensus now on what features people would
like to see in this simulator, while we are still designing the
framework. It is for certain that the simulator will not meet
everyone's wish-list, if we want a prototype to run in the next few
months, instead of the next few years. But we'd like to hear
suggestions/needs/expectations to help our design effort, and,
whenever feasible, to add proper hooks so that later on people can
easily plug in whatever pieces we do not provide.
Lixia Zhang (Lixia@XX.LCS.MIT.EDU)
MITRE CORPORATION
-----------------
No report received.
NTA & NDRE
----------
No report received.
ROCKWELL
--------
No input to report this month.
John Jubin (Jubin@A.ISI.EDU)
SRI
---
No report received.
UCL
---
The ISO Developement environment has been adapted to run over X.25.
We now have the thin TP0 layer over both TCP and X.25, and so will
be in a position to transparently relay ISO applications over a
full stack of ASN1, full ISO session and TP0, between PDNs and the
internet.
We have collated various inputs to the design of a TCP performance
measurement and analysis tool. These are being added to existing
Westine [Page 6]
Internet Monthly Report May 1987
features. Have already investigated a local problem with a
particular TCP and slow subnet routers with the tool.
We are looking at implementing a loop resolution and load sharing
scheme for multiply interconnected LANs using MAC level bridges.
Any input on bridge-bridge protocols would be interesting. The
problem of host address acquisition and cacheing in bridges when
hosts move is being looked at.
Jon Crowcroft (jon@CS.UCL.AC.UK)
UNIVERSITY OF DELAWARE
----------------------
1. Development continues on the Dissimilar Gateway Protocol
(DGP). Several details of the routing algorithm and data-base
management procedures were worked out with Linkabit staff.
Design of a developmental prototype for a three-Sunstation
network has begun.
2. An elusive bug reported last month in the new software
distribution for the fuzzballs has apparently been squashed
and the fuzzware installed in all machines at UDEL, Maryland,
Michigan and Rice, as well as Linkabit, Ford and the NSFNET
Backbone machines. This version more than doubles the
available buffer storage (11/73 configurations), improves
performance under congested conditions and fixes several minor
bugs.
3. A close look at NSFNET Backbone statistics a week after the
new fuzzware was installed showed that congestion is much
reduced, with the rate of packet loss dropping to less than
0.1 percent at the busiest gateway (PSC), in spite of its
current load of about six megapackets per week.
4. Yet a newer fuzzware version is now under test on UDel and
Linkabit machines. This version provides more efficient use of
buffer space and also removes certain obstacles hindering
expansion of the routing tables.
5. Several problems affecting primary/backup routing paths
between NSF and DCA swamps were identified, but not yet
completely resolved. The hardest to resolve may be
inconsistencies in the various copies of the manually
distributed data base used by the Unix gated routing daemon.
This is a problem in configuration management and control, not
a protocol issue.
Westine [Page 7]
Internet Monthly Report May 1987
6. A severe timewarp occured when an unlikely combination of
events disabled access to three out of four primary
(WWVB/GOES) radio clocks and one of three secondary (WWV)
radio clocks. Certain Internet hosts normally synchronized
directly to these clocks and providing time service to other
hosts synchronize to each other using the Network Time
Protocol (NTP). During the timewarp, which lasted for most of
a weekend, the most often used time servers had cranked
through four levels of NTP backup and locked on a dinky WWV
clock sitting on my desk at home. It is unlikely that the many
clockwatchers chiming these servers noticed the warp, unless
they were timing flies down to precisions in the order of a
few milliseconds.
7. Partly as the result of the timewarp, effort began on a plan
for a regularized Internet time service using designated
resources and access paths.
Dave Mills (Mills@UDEL.EDU)
NSF NETWORKING
--------------
UCAR/BBN LABS NSF NETWORK SERVICE CENTER (NNSC)
The NNSC hosted the second meeting of the NSFNET Federation of
Regional Networks on May 11 at BBN. Representatives from several
regional networks reported on the status of their networks. The
topics that the group discussed included committee activities,
funding for the continuation and expansion of the regional/state
networks, networking needs of the U.S. research community, and
NSFNET's role in the Internet community. The next Federation
meeting is scheduled for early September in New York.
On May 12, following the Federation meeting, the NNSC conducted a
Regional Network Managers' Forum on the topic of running network
information and operations center. Representatives from the
following organizations gave presentations: NSF, UCAR, NIC, BITNIC,
CSNET CIC, NNSC, JVNC, BBNCC. The session concluded with a tour of
the BBN Communications Company Network Operations Center.
The last of the four RFC's specifying the High-Level Entity
Management System (HEMS) is now nearly complete and circulating for
comment. In addition, the NNSC has been receiving an increasing
number of technical calls asking for assistance tracking local
network problems.
By Karen Roubicek (roubicek@nnsc.nsf.net)
Westine [Page 8]
Internet Monthly Report May 1987
NSFNET BACKBONE SITES
CORNELL UNIVERSITY THEORY CENTER
Backbone Operations
31,150,000 packets were delivered by the backbone in April, up 47%
from March. Thus in April we did better than a megapacket per day,
and recent measurements show we are doing about ten megapackets per
week (delivered to Ethernets). We are quite concerned about the
potential for congestion -- even though the fuzzballs still have
plenty of life in them the traffic is increasing even more quickly
than we anticipated. As always, more detailed reports are
available.
New software has been deployed in the fuzzballs which Dave Mills
will report on elsewhere.
We are seeing a lot of packets being sent into the NSFNET backbone
as a last resort, which the backbone just drops. This sort of
thing is very bursty. At the moment we are notifying the offending
gateway site and asking them to track down the culprits. Automatic
quenching of such bursts is being considered.
Coordination and Interoperability Issues
Gatedaemon
The plans for "gated" call for a new release to be available in
about 2-3 weeks. This version will have performance enhancements,
all known bugs fixed, and a few more options in regards to
filtering of routing information.
Specifically, the code for the "gated" program was cleaned up and
made more consistent. Unnecessary and redundant code was folded
into tighter code. This decreased the size (in lines of code)
considerably. Also, the coding style in regards to paragraphing
and commenting was made consistent throughout the program.
Westine [Page 9]
Internet Monthly Report May 1987
Enhancements, fixes include:
- - Full split horizon is implemented.
- - The interface code is much more robust. Changes can be made
to interface flags, metrics, even addresses without having to
restart gated. Interfaces are checked every 30 seconds
for changes.
- - PTP links are dealt with more consistently. The kludge
in dealing with SLIPnet interfaces has been removed and all
PTP's are handled the same way.
- - syslog is used throughout to report errors, notices,
warnings, etc.
- - There is a hold down of 2 minutes for each of the routing
protocols.
- - RIP now listens to host routes. This is necessary to deal
with PTP's more consistently.
- - There is now one table for host routes and one table for
net routes. The size of the hash table has been increased
and the hashing function has been changed to match the
Berkeley Kernel.
Enhancements planned before the release:
- - Filtering on source for routing information. The option
will exist to listen to given source addresses for named
networks. Example: only listen to information about default
from a.b.c.d and a.b.f.a
- - Support for SunOS 3.3 and VAX VMS. VAX VMS mods were
provided by David L. Kashtan.
- - The notion of subnet interfaces will be supported. This will
mainly be used to control where subnet information will be
supplied by RIP.
- - sending gated a signal will cause it to dump all it knows
into a file suitable for viewing.
Contemplations which may be in next release:
- - route history. The last route to the destination network
from each qof the supported protocols will be saved.
By Scott Brim (swb@devvax.tn.cornell.edu)
Westine [Page 10]
Internet Monthly Report May 1987
UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN
Connections to Argonne National Labs and University of Illinois-
Chicago are now in place. UIC has been allocated a net and will be
advertised from the Internet first week in June. The ANL
connection is a bit tricky since they have a MILNET connection on
that same Ethernet. The advertisment will be gated Hello'ing to
the UIUC fuzzball thence to PSC gateway.
We have converted the hyperchannel frontending the Cray to a subnet
of the connected the campus network. Therefore, the Cray is now
reachable and usable from the Internet. Currently running (and all
that's currently planned) is a CTSS Telnet server, FTP server, and
FTP client. As previously stated the Telnet server accepts line
mode only. Also, BSD4.3 users should change there echo control
character from ^E (which is an execptionally useful monitor escape
character to CTSS). Documentation should be available in a month
or so. In that time frame we will also register the implementation
with ISI.
The Vita-Link installation had been expected to be completed by 15
May. We are now expecting completion by 5 June. At that time UIUC
will be a hub (similiar to USAN) with Indiana University as its
first rim site. University of Chicago will follow this summer.
By Ed Krol (krol@uxc.cso.uiuc.edu)
JOHN VON NEUMANN SUPERCOMPUTING CENTER
EVENTS OF THE MONTH:
ROUTING:
The "gated" daemon is running and providing us with the tools to
interoperate between the JVNC and the NSFNET using the "hello"
protocols, JVNC and the Consortium, using the "rip" protocols, and
JVNC and the NRAC group, using "egp". We have found certain bugs,
that will be fixed by the new realease of the software from Mark
Fedor ( Cornell ). Some of the *new* features of the gated program
are the use of the "split horizon" and "hold down" techniques, to
avoid routing loops.
Sergio Heker, participated on the Internet Engineering Task Force
meeting at BBN.
MONITORING:
There is a current effort to finish phase I of the JVNCNET
monitoring system, the system will be robust and flexible enough to
be used in any network configuration and to provide the network
manager of that network with the tools to determine network
Westine [Page 11]
Internet Monthly Report May 1987
reachability, host reachability, links status, traffic flow, input
and output errors.
SATELLITE:
The TransLAN bridges were crashing very often and after rebooting
would come back on line. The problem seemed to be in the software
version that we were running. After updating the software they
seem to start working much more reliably. So far and for an entire
week there have no more crashes. The initial configuration that
VITALINK sent us was not right, therefore it took a couple of hours
to reconfigure the TransLAN's at JVNC, University of Arizona and
University of Colorado.
T1 LINES:
In general they are very reliable, and they don't constitute any
limitation in bandwidth to us. The main limitation is given at
this point by the routers that we use that cannot be driven to full
T1 rate.
We had a problem with the T1 line between JVNC and MIT, the problem
was detected to be at the Central Office of the Cambridge phone
company. When recognized by them, the problem was fixed. It took
a couple of days for this problem to get fixed, in the meantime MIT
as well as HARVARD and Brown couldn't get to JVNC or its attached
networks.
GATEWAYS:
JVNCE was put off line, and its line to Penn State University was
moved to JVNCB. The reason of the move is to use JVNCE for
supercomputer tests. The move was done during regular PM (note
that our routers have regular PM every monday morning since they
are VAX750s).
SUPER, the vax750's gateway to the University of Pennsylvania, was
down many times over the month due to disk problems.
Of the 18 gateways (not including jvnca), they were available
(reachable via JVNCnet lines) 95% of the time.
We received many power hits over the month, each one took the
entire network down. We recovered in any case in record time.
The Ungermann Routers that we are using to connect to the NRAC
schools seem to respond well to routing changes (using EGP), as
well as to the load that they have to deal with. Currently they
are connected with 56k lines, and the traffic on that lines is not
too high.
Westine [Page 12]
Internet Monthly Report May 1987
PSN:
The PSN is in place, the CSU/DSU's that connect JVNC'PSN with the
ones at the University of Maryland, Harvard University are tested
as well as the lines, the line to the University of Colorado hasn't
been declared up yet. The ACC board has been installed in JVNCA,
and tested. We are now waiting for DCA to come and connect the PSN
to our vax.
FUZZBALL:
The fuzzball's software has been updated by Hans-Werner Braun (U.
Michigan/ Merit network). The new software has much more buffer
capabilities than the previous version, and will alleviate the
existing congestion problem on the NSFNET backbone.
LAN:
The local Ethernets are very congested, and we are suffering from a
very high number of collisions. The traffic patterns seem to be
small "telnet" type packets, and a few very large packets (probably
ftp's of big files, and/or x-windows traffic) in burts. The
problem could be alleviated by adjusting dynamically the MTU's of
our interfaces, but we believe our Ethernets are not capable of
sustaining this kind of traffic anyway. A tenth of our traffic
comes from the NSFNET backbone gateway. This represents about 23%
of the NSFNET traffic going to each NSFNET backbone gateway.
MEETINGS:
We hosted a meeting of the Network Managers of the members of the
JVNC Consortium, at JVNC. The meeting's main subject was routers,
models of trust and autonomy. Recommendations were made to the
JVNC Network Technical Advisory Committee.
* The JVNC Consortium is formed by the following institutions:
Princeton University, MIT, Harvard, Brown, University of
Pennsylvania, Institute of Advanced Studies, Rutgers, Columbia,
University of Rochester, NYU and Penn State.
** The NRAC group is formed by: New Jersey Institute of Technology,
University Medical and Dental of New Jersey, and Stevens Intitute
of Technology.
For more information contact "heker@jvnca.csc.org"
By Sergio Heker
Westine [Page 13]
Internet Monthly Report May 1987
NATIONAL CENTER FOR ATMOSPHERIC RESEARCH AND UNIVERSITY SATELLITE
NETWORK PROJECT
The Institute for Naval Oceonography (INO) in Mississippi and the
Naval Research Lab (NRL)in Maryland will become members of the USAN
network in the near future.
A Sun 3 has now been installed at NCAR to be used to interface the
128.116 and 128.117 networks to the ARPANET and to provide a
gateway to the NCAR Crays. It will also provide a backup gateway
between 128.116 and 128.117. We intend to install GATED when it
becomes available for Sun OS 3.3.
All USAN sites except Miami and Wisconsin have isolated their local
networks behind gateways. These two sites should be behind gateways
by late summer. Isolating local traffic at USAN sites frees
satellite bandwidth used by USAN sites and lessens the load on
128.116, on which about 20% of the traffic is pass-thru (mainly
to/from NSFNET).
By Don Morris (morris@scdsw1.ucar.edu)
PITTSBURGH SUPERCOMPUTING CENTER
During May the PSC Fuzzball had little unscheduled downtime
compared to previous months. Line stability has also improved with
the new software versions brought up during the month, improving
buffering and probably fixing the UDP error conditions. The PSC
Fuzzball has remained at or near the top of the traffic throughput
list.
PSC-GW has continued to serve as the primary ARPANET-NSFNET
gateway.
On May 22 the number for our backbone ethernet between PSC-Gateway,
the Fuzzball and the Proteon Gateway serving our local net changed
from a Class C number (192.5.146.x) to a subnet of our Class B
number (128.182.1.x). This will help to relieve the tension on the
number of nets being advertised to the core.
Plans are progressing well for the installation of PSCNET. In
early June we expect to have T1 connectivity through Proteon
Gateways to the University of Maryland.
By David O'Leary PSC Communications
SAN DIEGO SUPERCOMPUTER CENTER
On Tuesday, 2 June, we will complete the installation of the
SoftwareTools mail system mentioned last month. This will give
SDSC a mail gateway between all connected networks; SMTP based plus
Westine [Page 14]
Internet Monthly Report May 1987
SPAN, HEPNET, BITNET and SDSCNET/MFENET. We have been exercising it
for several weeks and all seems in great shape. With this change,
mail from SDSCNET will no longer use the NMFECC gateways to the
Internet or BITNET. The correct gateways are now @SDSC for BITNET
and @SDS.SDSC.EDU for the Internet.
The latter reflects the change at SRI-NIC of our offical names.
This change took effect on 29 May. Our machines are now registered
as follows (Note: our new Nicknames were our previous Offical
names):
Address Offical Name Nickname(s)
192.12.207.20 luac.sdsc.edu sdsc-luac.arpa
192.12.207.21 a.sdsc.edu sdsc.arpa,sdsc-a.arpa
192.12.207.22 sds.sdsc.edu sdsc-sds.arpa
Where a.sdsc.edu is the path to our Cray X-MP/48.
Our 56k UC Office of the President line to UCLA to reach IMP #1 is
not yet inplace. The delay seems to be due to the placing, by UCOP,
of an IDNX node at SDSC. This will integrate our UC system
communications directly into the UC system T1 service and eliminate
6 of our 7 current cross campus tail circuits, new ones in the
future (such as for the line to IMP #1), and allow for rerouting
around bad ports, etc with much greater ease. The IDNX is expected
20 June. This service should also bring a Proteon to Proteon link
between SDSC and UCB.
Other lines expected during June are: 56k to NSI at NASA/AMES, 56k
to the SPAN Router at JPL (replacing an existing 9.6k), 9.6 SPAN
line to S-CUBED (a NASA contractor) in San Diego, and, at last, our
T1 Microwave link to the Salk Institute in San Diego. This last
line will be Proteon to Proteon and support both IP and DECNET
traffic.
By Paul Love (loveep@sds.sdsc.edu)
NSFNET REGIONAL AFFILIATED AND CONSORTIUM NETWORKS
BARRNET
All of our 6 nodes are "up" and sending packets, although
performance problems still plague at least two sites. We had a
minor routing crises when a Milnet gateway route got propagated
through all RIP-controlled hosts in the network and this caused a
collapse in ARPANET access at UCB. Release 7.3 of the Proteon GW
software cured this particular problem, but we still need a more
generalized "route filtering" algorithm in order to insure better
protection from routing table corruption (intentional or
accidental). We are continuing our seemingly never ending test of
T1 CSU's. It's clear so far that what CSU is best for a leased-
Westine [Page 15]
Internet Monthly Report May 1987
carrier circuit is not necessarily best for a privately owned
microwave system. It also takes a knowledegable technician to
understand all the issuses (e.g. which signaling technique is best
for your configuration: 1) Alternate Mark Inversion, 2) Binary
Eight-bit Zero Suppression, or 3) Zero Byte Time Slot Interchange ?
Hint: there's no one right answer). We plan to continue testing
and performance analysis during this next month. We still are not
sure when BARRNET will get connected to the NSFNET backbone, but
many of our users are already looking forward to using this route
for access to the supercomputer centers.
By Tom Ferrin (tef@cgl.ucsf.edu)
JVNCNET (Refer to JVNNSC backbone report)
MERIT
The initial implementation of the UDP interface to MTS (the
Michigan Terminal System), which is the operating system run by
several of our major mainframes, was completed this month. Work is
progressing on the generalized TCP/UDP module within the network.
Progress is being made on a TCP/IP driver for the Interlan board,
which will allow TCP/IP-over-Ethernet connections into our Unibus-
based Primary Communications Processors (PCPs). Currently our PCPs
support only our own internodal protocol over Ethernets, although
our Secondary Communications Processors have supported IP over
DEQNA-attached Ethernets for some time now.
We are continuing to make changes in our implementation of the IP
switching function to enhance reliability and congestion control.
On the public-relations side, we have made a presentation about
TCP/IP within Merit and the implications of that service to the
board of directors of MICIS, the Michigan Interuniversity
Consortium for Information Sharing; we have had the University of
Michigan's Vice-Provost for Information Technology use PC/IP
successfully; we have published information in our newsletter about
Merit's IP service; and we have reviewed a set of articles on
TCP/IP and PC/IP which were published in the University of
Michigan's computing newsletter in the first week of June.
Subnet routing deficiencies in Proteon gateways have forced us to
expedite the Merit ability to tunnel addresses and number ranges
through the Merit packet switching nodes. This became necessary
because, contrary to the Proteon documentation, their gateways do
not support a bit mask for subnets at this time. Proteon has
promised to fix this, but this situation is holding us up for some
installations on the UMich campus. Using the Merit nodes instead
worked nicely; Merit even allows us to associate a specific bit
mask for ANY routing entry. Without individual bit masks for
Westine [Page 16]
Internet Monthly Report May 1987
routing entries the Proteon gateways will still not satisfy all our
needs, even after Proteon implements their current plans for subnet
improvements. But neither would most of the other available
routers. It currently looks as if only Fuzzballs, the BRL Gateway
and Merit nodes have the desired subnet flexibility necessary for
more complicated routing topologies.
Hans-Werner Braun chaired a meeting of the Technical Committee of
the NSF Network Program Advisory Group on 12 May 1987. The agenda
for this meeting, which was also attended by some people from DARPA
and NASA, was to have some general discussions about what
technologies are available for higher speed backbones. Hans-Werner
Braun also attended Vint Cerf's IRI Engineering Task Force meeting
in Virginia on the 13-15 May 1987.
By Christine Wendt (Christine_Wendt@um.cc.umich.edu)
MIDNET
All of our node hardware has arrived and has been tested. A three
node network has been configured for test purposes at the Univesity
of Nebraska-Lincoln and is currently being stressed. Some of our
telephone lines have been installed; all of the lines should be in
by mid-June.
MIDNET will hold a two and one half day meeting on June 22, 23, and
24 in Lincoln, Nebraska. The meeting will consist of two parallel
sessions: technical and general. The technical sessions will
basically be how to fly a gateway router and operate a network
node. These sessions will be hands-on; all eleven nodes will be
connected and running in a large classroom. After the meeting,
each institution will take their node gear back to their campus and
connect to the MIDNET telephone lines and their campus network.
The general sessions will include updates on the various networks
in the internet, a tutorial on internet concepts, a supercomputer
site update, information on supporting supercomputer and network
users, information on MIDNET network support services, campus
connection strategies, and miscellaneous consortium business.
Questions about the meeting should be addressed to Doug Gale
(402/472-5108) at DOUG@UNLCDC3.BITNET, Mark Meyer (402/472-5434) at
MARK@UNLCDC3.BITNET, or Martyne Hallgren (402/472-5435) at
MARTYNE@UNLCDC3.BITNET.
By Doug Gale (doug%unlcdc3.bitnet@wiscvm.wisc.edu)
NORTHWESTNET
We still have not yet received the NSF grant for NorthWestNet. In
the interim, we have sent out an RFP for internetwork routers,
connecting the 11 planned network nodes. The participating
Westine [Page 17]
Internet Monthly Report May 1987
universities are:
Montana State University (Bozeman)
North Dakota State University (Fargo, for the ND State system)
Oregon Graduate Center (Portland)
Oregon State University (Corvallis)
University of Alaska (Fairbanks)
University of Idaho (Moscow)
University of Oregon (Eugene)
University of Washington (Seattle)
Washington State University (Pullman)
By Hellmut Golde (Golde%UWACDC.bitnet@WISCVM.WISC.EDU)
NYSERNET
As of 1 May 1987, NYSERNET has the following topology with 56kbit
links and Proteon Gateways. Reachability to NSFNET, ARPANET,
MILNET, etc is available to each site. May additions include SUNY
Buffalo, SUNY Binghamton and Syracuse University.
Clarkson
Syracuse--+ |
| |
Rochester--------Cornell---------RPI---Albany
| | |
Buffalo | |
| | |
Binghamton | |
| |
NYTEL----Columbia------NYU
| |
| |
| NYNEX S&T
|
BNL
CSNET Phonenet cutovers to peer-to-peer INTERNET mail follow 2 to 8
weeks after network connection. T1 cutover for NYU to Columbia and
Columbia to Cornell is schedule for July. The root domain server
is being tested internally to be released to the NYSERNET community
by the end of June.
By Marty Schoffstall (schoff@nic.nyser.net)
Westine [Page 18]
Internet Monthly Report May 1987
SESQUINET
As of June 1st, we are up with connections to two campuses:
Baylor College of Medicine, and
Rice University.
At each of these campuses is a cisco gateway. The gateways are
connected via a 56-kb/s link. This connection is very recent, so
any usage information would be meaningless.
During the month of June, we plan to install connections to four
additional campuses:
Houston Area Research Center
Texas A&M University
Texas Southern University, and the
University of Houston.
Negotiations are underway for establishing a direct connection from
SesquiNet to a site of the NSFnet backbone.
By Guy Almes, (almes@rice.edu)
SURANET
The following nets are being EGP advertised to the core on
SURANET's behalf.
128.61. Georgia Tech
128.109. TUCC
128.150. NSF
128.154. NASA Goddard
128.163. U of Kentucky
128.164. George Washington Univ
128.167. SURANET
128.169. U of Tennessee
128.173. Virginia Tech
192.5.57. Univ of Delaware (udel-cc)
192.5.219. Clemson
192.16.177. Univ of Alabama
It is currently expected that the T1 connection between SURANET and
the Pittsburgh Supercomputing Center will be installed and working
by the middle of June.
The 56kb phone line to Florida State University should be tested
and operational on or before June 25.
Discussions are continuing with several Federal Research
Laboratories about establishing connections to SURANET.
Westine [Page 19]
Internet Monthly Report May 1987
Telecommunication vendors for the Phase II SURAnet sites will be
selected and letters of intent issued during the first week in
June. Formal contracts will be executed shortly thereafter.
By Jack Hahn (hahn%umdc.bitnet@wiscvm.wisc.edu)
WESTNET
1. Our award for FY'87 is being processed by NSF. It is not clear
whether the funding will be sufficient to cover just the
hardware costs for IP Gateways and DSU/CSUs, but we hope so.
2. Our plan is to request cost sharing from Mountain Bell and
AT&T for circuit costs between September and the time we
receive our FY'88 money from NSF. New Mexico Technet has
already agreed to do so within the state of New Mexico.
3. The RFP for IP Gateways is scheduled to be mailed out from
CSU's purchasing office on June 4, with responses due July 8,
1987. It is anticipated that a vendor will be selected by July
18, 1987, and that IP Gateways will be delivered by mid
August.
4. Colorado Supernet is negotiating with NCAR and JVNC for access
to NSFNET. Currently, we are not being allowed access to
NSFnet by either site. Dave Wood is negotiating with NCAR and
JVNC to establish access. This is particularly critical in
that a number of researchers need to Telnet and FTP to sites
other than NCAR and JVNC. As the remaining Westnet campuses
come upon Westnet, this will become even more critical.
By Pat Burns (pburns%csugreen.bitnet@wiscvm.wisc.edu)
Westine [Page 20]
Internet Monthly Report May 1987
TASK FORCE REPORTS
------------------
APPLICATIONS -- USER INTERFACE
No report received.
AUTONOMOUS NETWORKS
No news to report.
Deborah Estrin (Estrin@USC-CSE.USC.EDU)
END-TO-END SERVICES
No report received.
INTERNET ARCHITECTURE
The INARC list was quiet this month. There are several issues
ripening on the plate; however, suggesting a meeting should be
scheduled in the near future. As funding problems seem to
have been resolved here, this chairman at least can get back
on the airplanes and resume conking other crania on these
issues.
Dave Mills (Mills@UDEL.EDU)
INTERNET ENGINEERING
No report received.
IRI ENGINEERING
This task force held its initial meeting May 13-15 at the
Corporation for National Research Initiatives (NRI) in Reston,
VA. In attendance for all or part of the meeting were Dave
Clark/MIT, Milo Medin/NASA, Steve Goldstein/NASA, Mike
Corrigan/OSD, Russ Mundy/DCA-DDN, Bob Braden/ISI, Jon
Postel/ISI, Phill Gross/Mitre, Hans-Werner Braun/UMICH, Scott
Brim/Cornell, Mike St. Johns/DCA, Dennis Perry/DARPA and Vint
Cerf/NRI.
The primary goal of this task force is to formulate and
recommend steps to stretch the utility of the TCP/IP protocol
Westine [Page 21]
Internet Monthly Report May 1987
suite while new designs are developed for much higher speed
communication regimes. Any alterations or additions to the
TCP/IP protocol suite that are recommended must maintain
backward interoperability with current implementations.
The group identified the need to revise the model of the
Internet to account for the multiplicity of administrations
(DARPA, DOE, NASA, NSF, DCA...), introduction of commercial
components (LANs, gateways, protocol implementations...),
improved performance, accommodation of more dynamics in the
mobile parts of the Internet, renewed attention to security
concerns both from the Defense point of view and from the view
of inter-organizational networking.
A report is in preparation summarizing the results of the
meeting and will emerge in the form of an RFC after
appropriate review by the task force participants.
Vint Cerf (Cerf@A.ISI.EDU)
PRIVACY
Nothing to report this month.
John Linn (Linn@CCY.BBN.COM)
ROBUSTNESS AND SURVIVABILITY
Nothing to report this month.
Jim Mathis (Mathis@tsca.istc.sri.com)
SCIENTIFIC COMPUTING
No report received.
SECURITY
No report received.
Westine [Page 22]
Internet Monthly Report May 1987
TACTICAL INTERNET
No report received.
TESTING AND EVALUATION
No report received.
Westine [Page 23]