<NIS.NSF.NET> [IMR] IMR89-02.TXT
FEBRUARY 1989
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).
TASK FORCE REPORTS
------------------
APPLICATIONS -- USER INTERFACE
No report received.
AUTONOMOUS NETWORKS
The ANTF held a productive meeting February 14-16 at ISI.
During the first two days we were joined by several members of
the IETF ORWG. We spent much of the first day articulating the
context in which policy based routing will be implemented and
the problems that it needs to solve. The second day was spent
detailing our agendas and discussing several of the individual
projects specified (see upcoming ANTF meeting report for
details). In addition we discussed two particular routing
Westine [Page 1]
Internet Monthly Report February 1989
proposals, one by Dave Clark on Policy based routing for the
Internet (soon to be published as an RFC), and one by Greg Finn
on Cartesian Routing (available thru ISI). The third day of the
meeting was held in conjunction with the Privacy and Security
Task Force. During the first half of the day we addressed
labeling and security in policy routing protocols. The second
half of the day addressed the more general topic of denial of
service attacks on routing protocols.
A more detailed report on the meeting will be available shortly
from me (greatly assisted by the meeting notes taken by Sharon
Anderson).
Deborah Estrin (Estrin@OBERON.USC.EDU)
END-TO-END SERVICES
VMTP:
o The Unix kernel implementation of VMTP was released to the
world November 1988. Several fixes and extensions have
been added since.
o There will be an article on VMTP in an upcoming issue of
the IEEE Communication magazine, in an edition on high-
performance protocols.
o A version of X11 Mazewars is running at Stanford using VMTP
and multicast. They are still modifying it, but hope to
release it in the next few months.
o Dave Cheriton has presented VMTP to ANSI committee X3S3.3,
proposing it as a possible standard.
IP Multicasting:
o Steve Deering has been debugging and improving the RIP-
based multicast router developed at BBN and described in
RFC-1075. He expects to release it very shortly.
o Berkeley has not yet released the IP multicast host code;
it is currently scheduled for BSD 4.4.
o Craig Partridge and others at BBN are writing up a
comparison of the distance-vector-based and SPF-based
multicast routing protocols done at BBN.
Westine [Page 2]
Internet Monthly Report February 1989
o Discussions are underway with the Open SPF IGP (OSPFIGP)
Working Group of the IETF, concerning the possbible
inclusion of SPF-based routing in their protocol.
Bob Braden (Braden@ISI.EDU)
INTERNET ARCHITECTURE
At the December IAB workshop and meeting in Santa Clara, it was
suggested that INARC take up the issue of the limits of the
Internet architecture and protocols in the light of present and
projected future requirements. A meeting has been proposed for
the May/June frame at a date and place to be decided. The
meeting is to be organized as a workshop similar to the one held
a year ago. Participants will be encouraged to present a
prepared talk and/or document, but this is not a requirement for
admission. While participants may be invited on the basis of
their known expertise, biases and vocalities on the issues,
participants outside the IAB and its dependencies are actively
encouraged. Following the workshop the INARC will meet
specifically to discuss issues raised and to determine future
study plans and assignments.
Suggestions for agenda items and meeting date and place, as well
as intentions to attend, can be sent to the chair at
mills@udel.edu. The meeting date and place will be announced in
the INARC report for next month.
Dave Mills (Mills@HUEY.UDEL.EDU)
INTERNET ENGINEERING
1. The next meeting of the IETF is April 11-14 in Cocoa Beach
Florida. The meeting will be hosted by Kennedy Space
Flight Center, although the actual meeting site will be the
Cocoa Beach Hilton. KSC is planning an extended tour of
the Space Center facilities on Friday afternoon (April
14th).
2. As a direct response to discussions in the January IETF
Plenary session, the April meeting has been scheduled for
3.5 days. This will allow time for 2 full days of Working
Group sessions, 1 full day of technical presentations, and
.5 days for WG reports. The final agenda for the April
meeting is still being compiled.
Westine [Page 3]
Internet Monthly Report February 1989
3. Meeting information (eg, hotel, directions) has been mailed
to the IETF mailing list. For a copy, or additional
information, send a message to Karen Bowers (NRI,
bowers@sccgate.scc.com) or Sara Tietz (ACE,
sara@spam.istc.sri.com).
4. The dates and locations for the five IETF meetings
following April have been finalized. Please mark your
calendars! The dates and locations are:
July 25-28, 1989 Stanford University
October 31 - Nov 3, 1989 University of Hawaii
February 6-9, 1990 Florida State University
May 1-4, 1990 Pittsburgh Suppercomputer Center
July 31-August 3, 1990 University of Washington
Phill Gross (gross@sccgate.scc.com)
INTERNET MANAGEMENT
No report received.
PRIVACY
The IAB Privacy Task Force had a productive three-day meeting at
USC/ISI in Marina del Rey, California on 14, 15, and 16 February
1989. Privacy task force attendees were: Dave Balenson (now
representing TIS), Matt Bishop, Morrie Gasser, Russ Housley,
Steve Kent, John Linn, Jim Nechvatal (new NIST representative),
Dan Nessett, Rob Shirey, and Dave Solo.
The first two days were devoted to Privacy Task Force
discussion. We addressed final revision issues on the
electronic mail privacy enhancement RFCs, including a decision
to split off a new "Privacy Enhancement for Internet Electronic
Mail: Part III: Algorithms, Modes, and Identifiers" and to defer
detailed certificate ordering information to a later follow-on
RFC. Ralph Merkle of Xerox PARC gave a presentation on a family
of new encryption and hash algorthms which he has designed and
offered for consideration.
The session on the 16th was held as a joint meeting with the
Autonomous Networks Task Force, discussing policy routing issues
and other topics of mutual interest.
Westine [Page 4]
Internet Monthly Report February 1989
The next PTF meeting was tentatively scheduled for 23-25 May at
UCL, subject to reconfirmation with Steve Wilbur. Dartmouth
College is a possible venue for the following meeting, probably
in October.
John Linn (Linn@CCY.BBN.COM)
SCIENTIFIC REQUIREMENTS
No report received.
DSAB
----
No report received.
Westine [Page 5]
Internet Monthly Report February 1989
BOLT BERANEK AND NEWMAN INC.
----------------------------
WIDEBAND NETWORK
A message describing transition plans for the Wideband Network was
distributed on February 15 by Mark Pullen of DARPA/ISTO. It is
included below for the benefit of the Internet research community.
TO: Widebandnet Sponsors and Users
FROM: DARPA/ISTO
SUBJECT: Transition of Widebandnet to Fiber Backbone
As described in my previous message of 12 July 1988, the
satellite carrier's contract for Widebandnet is at an end and
the net is being transitioned to a fiber-optic T1 backbone, as
the first operational component of the Defense Research
Internet. It is recommended that users of Widebandnet plan
minimum reliance on the net during the months of March and
April, 1989.
The Widebandnet transition is being coordinated with the
installation of a number of cross-country terrestrial T1
circuits which will create a new National Networking Testbed
(NNT). The creation of the NNT is an initial step towards the
establishment of the Defense Research Internet (DRI). Among
other applications, NNT bandwidth will be used to provide
continued support for both the internetwork and ARPANET inter-
switch trunk (IST) traffic that is now carried by the
Widebandnet. The internetwork traffic will be serviced by a
newly designed terrestrial coast-to-coast network consisting of
high-speed Butterfly switches linked via a multidrop NNT T1
circuit. The ARPANET IST traffic will be serviced by 56 Kbit/s
trunks to be derived from a separate NNT T1 circuit.
The Widebandnet transition will commence on or shortly after 1
March 1989, when satellite channel access will be terminated for
four Wideband earth stations: M/A-COM Government Systems in San
Diego, CA; US Army CECOM in Ft. Monmouth, NJ; Carnegie Mellon
University in Pittsburgh, PA; and Ft. Huachuca, AZ. (The latter
two of these sites have not been active on the network for some
time.) Channel access for the six remaining earth stations (ISI,
Lincoln Laboratory, DCEC, SRI, RADC, and BBN) will be terminated
in the early April time frame. The exact date for final
Widebandnet operation will be a function of progress made in
transitioning the Wideband-based ISTs to the NNT. As the
Westine [Page 6]
Internet Monthly Report February 1989
transition approaches, cutover schedules will be provided.
The high-speed NNT-based network will ultimately provide a level
of service and functionality to its users exceeding that which
has been provided by the satellite-based net. Hardware/software
interfaces to the new network will initially be the same as
those currently used in Widebandnet and, other than a reduction
of network message delays due to the elimination of satellite-
channel propagation, users of the standard connectionless
Widebandnet "datagram" service should notice little change in
their wide-area networking support. Initial operation of the
terrestrial network is scheduled for the April time frame. By
1990 we expect to be offering various high-speed packet services
in the DRI. At that time the network interfaces are likely to
change.
Even with the best of planning, a change such as this
necessarily entails some amount of disruption and temporary loss
of service. We solicit the patience of current Widebandnet
users during this transition period as we move closer to the
achievement of the high-speed research network environment of
the future.
Mark Pullen
SATNET
SATNET performance was excellent during February. Statistics
collected by ISI showed an average of 100% uptime for the SIMPs and
99% uptime for the attached European gateways.
The SATNET connection to the UK (Goonhilly) has been terminated on
paper but at this time is still operating. It is expected the
connection will be turned off physically some time in March. This
will leave only the U.S. and Italy operating on the SATNET. The
new direct link to Italy is expected to be operational in April at
which time the SATNET will be completely turned off.
INTERNET R&D
The Internet continues to grow. At the end of February there were
over 640 networks in the Internet.
We have started working on deploying new Butterfly Gateways at
SHAPE Technical Center in The Hague, and at WPC in West Germany and
additional Butterfly Gateways at UCL and RSRE. These will be part
of the European DRI. We have also started working on deploying
three Butterfly Gateways for SURAN sites (two for Ft. Monmouth, one
Westine [Page 7]
Internet Monthly Report February 1989
for SRI).
The IETF Open Routing Working Group meet in February at ISI.
Bob Hinden (Hinden@BBN.COM)
CORNELL UNIVERSITY
------------------
Scott Brim has been working on projects in the IETF
Interconnectivity Working Group and the Autonomous Networks Task
Force (q.v.). Among other things he has been developing plans for
near-term steps on the way to full policy-based routing, working
out specifics for policy-based routing experiments with the RIGs,
and defining interactions between the components of a policy-routed
network, within and between administrative regions.
GATED: This month the gatedaemon underwent a lot of internal
reorganization for improved operation and as a large step toward
modularity and ease of adding new protocols. Interdependence
between protocols was reduced to the point where it should be
possible to select which protocols should be included at compile
time.
As part of being redesigned for modularity, gated's timing routines
have been rewritten to be more general and more accurate. This
will also prevent synchronization of routing updates on broadcast
protocols such as RIP (the "boxcar" effect).
Also rewritten were the gated tracing and logging facilities, now
combined into one routine which logs to the trace file and/or the
system log depending on arguments passed.
Gated's EGP2 implementation was completely rewritten to comply
explicitly with RFC904. The previous code deviated from the RFC in
many respects. The new implementation is a more formal
implementation of a state machine and should provide a good base
for EGP3. It has also provided us with a better understanding of
the EGP2 protocol and uncovered a couple of glitches in RFC 904.
This implementation has also uncovered a few bugs in the
Butterflies which have been reported to BBN. Testing has just
begun against Proteon's EGP implementations and we hope to set up a
test with other routers as well.
On gated's near-term list of things to do are:
* generalizing the event-driven update system.
Westine [Page 8]
Internet Monthly Report February 1989
* redoing the command syntax and parser to make adding newer
protocols and experimenting with policy-based routing
prototypes easier.
* EGP3 (as soon as the specifications are ready). Along
with this we will be reorganizing the routing database,
keeping multiple routes, and conceptually treating the system
kernel as another gateway to which we give selected
information.
* OSPFIGP.
* SNMP traps, configured per community and per interface, per
list of recipients.
Jeff Honig and Scott Brim (swb@chumley.tn.cornell.edu,
jch@devvax.tn.cornell.edu)
ISI
---
INTERNET CONCEPTS PROJECT
The basic IP congestion control algorithm assumes no new
functionality on the part of gateways. Simulation results indicate
that congestion is well controlled but resources are often unfairly
split between multiple competing sources. A modificationm that
assumes that gateways return source quenches for packets drawn at
random from their overflowing queues, rather than always returning
a source quench for the packet that caused the overflow, appears to
improve fairness. This is being studied.
Greg Finn
Jon Postel visited Digital Sound Corporation in Santa Barbara, to
discuss networks. Paul Mockapetris attended the FRICC White Pages
Working group Meeting, in Washington, D.C, February 6 and 7.
Six RFCs were published this month.
RFC 1088: McLaughlin, L., "A Standard for the Transmission of IP
Datagrams over NetBIOS Networks", TWG, February 1989.
RFC 1089: Schoffstall, M., C. Davin, M. Fedor, and J. Case
"SNMP over Ethernet", February 1989.
RFC 1090: Ullmann, R., "SMTP on X.25", February 1989.
Westine [Page 9]
Internet Monthly Report February 1989
RFC 1091: VanBokkelen, J., "Telnet Terminal-Type Option",
FTP Software, Inc., February 1989.
RFC 1092: Rekhter, J, "EGP and Policy Based Routing in the New
NSFNET Backbone", T. J. Watson Research Center,
February 1989.
RFC 1093: Braun, H. W., "The NSFNET Routing Architecture",
Merit, February 1089.
Ann Westine (Westine.ISI.EDU)
LOS NETTOS
Los Nettos is being used more and more as the path to the internet
for member sites. TIS is no longer advertising their net 10
address. USC and ISI are both making similar changes.
Most problems with Datatel 3552 CSU/DSU units are fixed with 1.2
firmware in the units. Two problems are still outstanding.
The T1 line for JPL is scheduled to be installed early March.
Walt Prue will represent Los Nettos at the FARNet meeting held at
Stanford March 1, and 2.
Plans and orders for Phase 2 of Los Nettos installations are
progressing.
Walter Prue (Prue@ISI.EDU)
MULTIMEDIA CONFERENCING PROJECT
The packet video host (PVP) was extended to enable interfacing with
the WIDCOM video codec.
Eve Schooler, Steve Casner, Dave Walden (schooler@ISI.EDU,
casner@ISI.EDU, djwalden@ISI.EDU)
NSFNET PROJECT
Considerable effort this month was devoted to submitting a renewal
proposal to NSF.
Annette DeSchon and Bob Braden attended a meeting of the Autonomous
Networks Task Force, and Bob Braden attended a meeting of the Open
Routing Working Group, both held at ISI.
Westine [Page 10]
Internet Monthly Report February 1989
Annette DeSchon started work on some additions to BFTP to support
FTP Restart. With Jon Postel, we are looking into a possible new
definition of the FTP Restart function that was proposed by Rick
Adams.
Bob Braden worked on updating the Host Requirements RFC with the
changes agreed to at the Austin meeting, plus hundreds of changes
proposd by Mike Karels and Charlie Lynn. This work is still in
progress.
Bob Braden and Annette DeSchon (Braden@ISI.EDU, DeSchon@ISI.EDU)
FAST PROJECT ON REMOTE ACCESS OF GRAPHICS DATABASES
Alan continued working on making the X version 11R2 server work
under Sun Overview (we are currently installing X11R3).
Alan is currently working on defining a Remote Execution protocol.
Alan Katz (Katz@ISI.EDU)
MIT-LCS
-------
Dave Clark, Karen Sollins, and Lixia Zhang attended Autonomous
Networks Task Force meeting held on February 14-15; Dave and Karen
attended joint ANTF and Privacy Task Force meeting on February 16;
Lixia attended Open Routing Working Group meeting on February 16-
17, all held at USC-ISI.
The Gateway Testbed project is moving steadily towards its goal of
deploying a national network for experimentation with gateway
algorithms and new networking concepts. Software to support the
gateway platforms is being readied. A decision on the bandwidth of
the Testbed links will pave the way for finalizing the budget and
subsequent deployment of equipment.
Lixia Zhang (Lixia@XX.LCS.MIT.EDU)
MITRE Corporation
-----------------
No report received.
Westine [Page 11]
Internet Monthly Report February 1989
NEARNET
-------
Installation of the hub of the New England Academic and Research
Network (NEARNet) was completed in February, connecting Boston
University, Harvard University, and the Massachusetts Institute of
Technology. NEARNet will eventually serve many New England
organizations engaged in research, development, and education. The
initial hub was a cooperative effort among the three participants.
Seed funding to expand the hub to a Phase 1 network of nine sites
in Greater Boston (including the three above) has been provided by
DARPA, with additional funding from the sites themselves. We
expect all nine to be operational in April, with others added soon
thereafter.
Seven of the Phase 1 sites will be linked at 10Mb rates using
microwave technology. The network will be redundant with automatic
load sharing handled by Cisco routers which are capable of routing
packets at Ethernet speeds. Outlying areas will be served by
microwave connections, T1 land lines, and 56Kb land lines. NEARNet
will be connected to the Defense Research Internet, the ARPANET,
and NSFNET.
Boston U., Harvard, and MIT are acting as the Steering Committee
for NEARNet initially; they have contracted with BBN to be the
facility operator. An expanded Planning Group, with
representatives from several New England states, will hold its
first meeting in April.
Jim Bruce, MIT
Steve Hall, Harvard
John Porter, Boston U.
for the NEARNet Steering Committee
by Daniel Long (long.sh.cs.net)
NTA-RE and NDRE
---------------
No report received.
Westine [Page 12]
Internet Monthly Report February 1989
SRI
---
SRI hosted the User Interface Taskforce meeting chaired by Keith
Lantz. As part of the meeting, the teleconferencing facility was
used to demonstrate the latest version of multimedia conferencing
software from BBN. We used this opportunity to experiment with
large screen displays for the computer screen information to
facilitate larger groups (about fifteen) in one room. Although
special lighting was used to accomodate both the video camera and
the large screen projector, improvements are still possible.
Earl Craighill (EJC@TSCA.ISTC.SRI.COM)
UCL
---
We held some informal discussions on mutual aid for Transatlantic
links between European and US Research Networks.
We have just upgraded the nodes on our FDDI network and expect it
to run as a service from Easter.
Both of these projects have unearthed some problems with DECNET
addressing as connectivity grows.
We have just joined another RACE (Research into Advanced
Communications for Europe) project called CAR (Communications for
the Automotive industry under RACE). They have some interesting
(stressful) applications for Internetworking, especially in the
area of Distributed CAD/CAM for car design.
John Crowcroft (jon@CS.UCL.AC.UK)
UNIVERSITY OF DELAWARE
----------------------
1. Our grads continue hacking theses and various odd jobs. Mike
Minnich is working on feedback-control models for networks of
hosts and gateways. Paul Schragger brought up the Profile
system for an experimental white-pages server and NYSERNET
SGMP for watching our gateway rascals. Jeff Simpson attended a
meeting of the Autonomous Networks Task Force and is now
sloshing in formal grammars for expressing routing policy. Dan
Grim and Chuck Cranor hexed an ancient VAX witch and exorcized
the last traces of our old class-C network numbers.
Westine [Page 13]
Internet Monthly Report February 1989
2. Mike Davis completed a particularly messy job of cataloging
and archiving precious and prodigous monitoring data collected
during the two-year bumpy life of the old NSFNET Fuzzbone
network. This includes hourly and weekly summaries of packet
counts by interface, as well as error messages and various
categories of dropped packets. Careful analysis showed that
the mean hop count across the network was somewhat larger than
expected (about 1.8 hops) for the particular topology used,
which suggests performance could have been considerably
improved by simply rehoming some of the links.
3. Documentation for the new Network Time Protocol (NTP) Version
2 was completed and stashed on louie.udel.edu for public
display. Two PostScript documents are now available:
pub/ntp/ntpx.ps, which describes the architecture, models and
algorithms, and pub/ntp.ps, which is the specification itself.
The new version, which is compatible with previous versions,
includes an authentication feature using DES crypto-checksums,
as well as improved clock- selection and local-clock
synchronization algorithms using an adaptive-parameter phase-
lock loop. Pending review of the authentication feature by the
Privacy Task Force, the specification, now a Draft Internet
Standard (DIS), will be proposed for promotion to Internet
Standard (IS).
4. A good deal of activity occurred this month testing the new
NTP Version 2 beta implementations for Fuzzball and Unix
systems. Various bugs were found and corrected in the
specification and implementations. A software simulator was
constructed to explore the synchronization behavior over
intervals up to several days. Marion Hakanson, Doug Kingston
and Jerry Aguirre were particularly helpful and patient in
finding subtle interactions with the Unix kernel that became
evident only after days of observation. The new Unix daemon
ntpd should be ready for public distribution early next month.
5. A high-stability, temperature-compensated crystal oscillator
was installed on the Backroom Fuzzball to test the ability of
the new NTP local-clock synchronization algorithm to maintain
a long-term frequency stability in the order of a millisecond
per day (about 10**-8) relative to NIST standards. The results
show that, while stabilities in the order of 10**-7 can be
achieved using presently available radio clocks and the
Fuzzball implementation, further improvements appear to be
limited due to the tracking loops in the radio clocks
themselves. However, further improvements may be possible as
inexpensive LORAN-C and GPS radiopositioning receivers become
available. Ideas for incorporating precise timekeeping
Westine [Page 14]
Internet Monthly Report February 1989
information as opportunities develop during the development of
these systems are being discussed informally with the U.S.
Coast Guard.
Dave Mills (Mills@UDEL.EDU)
Westine [Page 15]
Internet Monthly Report February 1989
NSF NETWORKING
--------------
NSF NETWORKING
UCAR/BOLT BERANEK AND NEWMAN INC., NNSC
Craig Partridge attended the IAB Workshop on White Papers in
Washington, D.C.
by Karen Roubicek (roubicek@nnsc.nsf.net)
NSFNET BACKBONE (MERIT)
New Topology
Merit and its partners, MCI and IBM announce a major redesign
project to improve connectivity and move NSFNET toward the
higher speeds required in the future. The new topology is
scheduled to begin implementation during second quarter 1989.
By increasing the number of T1 circuits in the backbone,
connectivity to spur sites will be improved and network
capacity as a whole will be increased with each node to have
multiple T1 circuits. Primary goals of the new topology are
to maximize network robustness and minimize bit traversal
delay. A detailed implementation schedule is nearing
completion and will be finalized in cooperation with the
NSFNET regional networks. Questions should be sent to:
NSFNET-info@Merit.edu
NSFNET BACKBONE
As of February 28, 1989, the NSFNET has completed eight months
of production. Traffic and network connections have continued
to increase during this period. The number of "allowed"
networks began at 173 primary and 30 secondary on 7/1/88 and,
as of 2/28/89, stands at 410 primary, 177 secondary, and 28
tertiary with new networks coming on line almost every day.
Westine [Page 16]
Internet Monthly Report February 1989
------------------------------------------------------------
Comparison of Packet Counts
January 1989/February 1989
Packets In Packets Out
January 467,753,653 497,787,338
February 573,390,288 593,669,705
% increase 22.6% 19.3%
------------------------------------------------------------
Even though February is a shorter month, packet counts showed
a marked increase over January 1989. Despite the usual drop
in traffic on weekends the upward trend was steady through the
month, peaking on the 23rd and 24th.
NNStat
February marked the beginning of the use of more sophisticated
methods for gathering statistics within the NSFNET backbone.
These data, which are being compiled continuously at all 13
backbone nodes, include packet counts, protocol frequencies,
packet size distributions, switching rates, and
source/destination network matrices. Information gathered
will aid the NSFNET partners in the analysis of network usage
and will provide needed input for the planning of network
configuration and expansion.
by Patricia G. Smith (Patricia_G._Smith@um.cc.umich.edu)
NSFNET BACKBONE SITES & MID-LEVEL NETWORK SITES
BARRNET
No report received.
CERFNET
Among notable activity are the receipt of signed subscription
agreements from the University of California at Los Angeles
(UCLA) and the University of California at Santa Barbara
(UCSB). Hughes have recently submitted a request for CERFnet
subscription. They will have a 56 kilobit line connection to
the California Institute of Technology and are scheduled for
April turn-up.
The cisco gateway at the San Diego Supercomputer Center was
rebooted on February 17th to initiate the link between UCLA
Westine [Page 17]
Internet Monthly Report February 1989
and Los Nettos. (Statistics will reflect this restart as they
are zeroed out). The following is a comparison of packets
switched.
---------------------------------------------------------
Packets In Packets Out
Start-January 3,248,796 1,414,807
February (2/1-2/16) 8,473,296 1,750,897
February (2/17-2/28) 5,076,406 2,315,318
February Total 13,549,704 4,066,215
% increase 317% 187%
--------------------------------------------------------
A substantial increase in packets processed is noted as the
network is becoming more active. The amount of packets
processed is expected to continue to rise as new sites come up
on the network.
The CERFnet pilot network is functioning well. A circuit
outage occured on February 17th from the San Diego State
University to the San Diego Supercomputer Center. The
connection was reportedly down from 9:00 a.m. until 13:30 p.m.
and the trouble was attributed to a Pacific Bell carrier
problem.
CERFnet will distribute the second issue of our bi-monthly
newsletter, the CERFnet News, starting March 6th.
by Karen Armstrong (armstrongk@sds.sdsc.edu)
CICNET
No report received.
UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN/NCSANET
No report received.
Westine [Page 18]
Internet Monthly Report February 1989
JVNCNET, NORTH EAST RESEARCH REGIONAL NETWORK
Monthly Status Overview:
Overall, the JvNCNet network has performed very well for the
month of February. The amount of traffic went up
significantly and steadily as seen below, the reachability to
the JvNCnet sites went down mostly due to a scheduled power
outage at JvNC as can be seen below.
The overall up-time for the network this month was 95.88%.
The "estimated" amount of traffic in and out of the JvNCnet
system for this month was 732,288,159 packets. This traffic
is a significant increase from last month's (657,982,496
packets). The traffic coming from the NSFNet was
approximately 28.80% of the total, the traffic coming from the
ARPANET amounted to approximately 9.25 %, the traffic from
NORDUnet was 1.99%, leaving 60.96% of traffic among JvNCnet
Regional sites. Of all the JvNCnet Regional sites, MIT
contributed with slightly more traffic than Rutgers. Rutgers
has been the single site on the network to contribute with
more traffic for the last few months.
The router for NORDUnet arrived ("bart") and was installed,
therefore the connection of NORDUnet was moved from Trillian
to Bart.
A scheduled power outage affected JvNC and all incoming
telecomunications services. This power outage was necessary
to put one of the Motor Generators (MG) in place. This
disruption will help the network equipment in the future by
isolating it from the electric power "glitches" produced
outside the Center.
A problem which affected six North East institutions was
finally confirmed and resolved. The problem was severe
congestion on the Cisco routers connecting: Boston University,
University of Massachusets, Dartmouth, Northeastern, Yale and
Wesleyan. The problem was actually due to a "bug" on some
EPROMs that were being used on those routers, in addition of
higher load on the network. A technician was dispatched from
JvNC to replace the EPROMs. The replacement took four days to
be accomplished. The results have been remarkably good.
Congestion drops have been reduced significantly on those
gateways.
All the Cisco routers have been upgraded with version
Westine [Page 19]
Internet Monthly Report February 1989
7.0(399). This software upgrade couldn't be completed until
the EPROMs in some routers were changed. After the EPROMs
were replaced, the routers had no problem to boot their new
image.
The circuit and equipment for the connection to JANET (Joint
Academic Network) in the UK to the NSFnet at JvNC has been
ordered. The router arrived and has been tested and
configured. The circuit will be installed (as per Contel)
before the end of March.
Siemens Research is now operational and will be incorporated
to the traffic report next month.
The installation of the first group of the JvNCnet Phase II is
in progress. We expect by the second week of March to see the
Newark Backbone Node, as well as the following end nodes will
become operational: Rutgers, NJIT, UMDNJ, Stevens Institute.
A few weeks later the Phase I/Rutgers circuit will be
disconnected. The next RCI/JVNC Implementation team will meet
on the seventh of March at JvNC.
The JvNC Network Operations Center (JvNCNet NOC) provides
support to the JvNCNet network as well as the internal network
of the JvNC. The NOC is staffed from 9am to 5pm Monday
through Friday. The computer operators monitor and perform
minor troubleshooting tasks between 5 p.m. and 9 a.m. and are
the backup for the Network Operations Center. Network Staff
are on call after hours and during weekends and holidays. The
JvNC Network Information Services Center is now staffed, and
provides support from 9am to 5pm Mondays to Fridays.
For more Information Contact:
JvNCnet NIC,
electronic mail address: "JvNCnet-nic@jvnca.csc.org"
phone number: (609) 520-2000, x387
by Sergio Heker (heker@jvnca.csc.org)
MERIT/UMNET
No report received.
MRNET
One new member, Honeywell, attached to MRNet during February.
ARPAnet connectivity was often poor, and Internet access in
Westine [Page 20]
Internet Monthly Report February 1989
general seemed to suffer as a result.
On March 1, the U of Minn's ARPAnet line was officially cut
(though still functioning). The effect on Internet access is
not yet clear, but it will simplify network operations.
In late February, CICnet lit up, promising substantially
faster access to other NSFnet sites for U of Minn. and MSC
users; other MRNet members are not yet using CICnet.
We ran NNStat for part of this month with some interesting results:
- our largest users are UofM/MSC and CDC;
- most-used TCP/UDP services are NNTP, rlogin, telnet,
SMTP, domain and FTP in about that order, with NNTP about
3 times FTP;
- inter-member traffic is 50-65% of all non-routing traffic
i.e. we're not all just talking across the Internet.
by Stuart Levy (slevy@msc.umn.edu), and Tim Salo
(tsj@ucmsc.umn.edu)
NATIONAL CENTER FOR ATMOSPHERIC RESEARCH AND UNIVERSITY
SATELLITE NETWORK PROJECT
The National Institute of Standards and Technology (NIST),
Boulder Division (net 132.163) is now gatewayed to the NSS via
the NCAR node. The Boulder NIST was formerly part of the
Gaithersburg NIST (129.6) and still maintains a direct link to
that net.
USAN statistics for February show the following daily average
traffic received by the USAN hub from the various sites.
Supervisory data between the hub and each site account for
over 0.1 Mbytes.
Naval Research Lab 4.53 Mbytes
University of Maryland 0.18 Mbytes
University of Miami 10.87 Mbytes
University of Wisconsin 12.25 Mbytes
Woods Hole 9.08 Mbytes
Oregon State University 2.95 Mbytes
Institute for Naval Res. 4.16 Mbytes
by Don Morris (morris@windom.ucar.edu)
Westine [Page 21]
Internet Monthly Report February 1989
NORTHWESTNET
We are presently in the process of a major effort to
understand and re-engineer the network. Some sites experience
unacceptably long packet latencies due to multiple hops on
19.2Kb voice grade circuits. Although our network manager
(Boeing Computer Services) has been unable to provide detailed
outage or traffic data, we don't seem to have bandwidth
problems -- despite a mix of 56Kb and 19.2Kb circuits (no T1
here in the boondocks!). A major emerging issue is the degree
of path redundancy desired/affordable.
Also, we have approved in principal a new class of network
memberships targetted at small for-profit research
organizations.
Apart from one significant line failure in late February, the
network has operated reliabily.
by JQ Johnson (jqj@hogg.cc.uoregon.edu)
NYSERNET
No report received.
OARNET
No report received.
PITTSBURGH SUPERCOMPUTING CENTER
PSC has installed and is now testing a new gateway which will
be the primary path between the NSFnet and the ARPAnet. The
current gateways at Cornell and UIUC are scheduled to be
disconnected from the ARPAnet by the end of March.
by Matt Mathis (mathis@fornax.ece.cmu.edu)
SAN DIEGO SUPERCOMPUTER CENTER
On Friday, 13 Jan., SDSC began peering with the Buttergates.
We are working with SRI/TGV to implement the FTP command PASV
in Multinet. This will support BFTP, among other things.
NFS is now in production use under Multinet on our VAX
Cluster.
Westine [Page 22]
Internet Monthly Report February 1989
During Feb. a T1 link to Naval Ocean Systems Center, Pt. Loma
[San Diego], Calif. became operational. This link uses a pair
of Bridge IB/3's together with a dual-Ethernet VAX 11/750 at
NOSC to tie the two locations together while separating the
two nets.
The CERFnet test network has been expanded with a 56k link to
UCLA provided by the University's Office of the President.
SDSC's old NSFnet I Fuzzball is providing Stratum 1 NTP
services.
The University of Nevada has selected SDSC as its connection
point into NSFnet for their statewide system. This system
will make use of an upgrade to an existing microwave system to
carry a data channel throught the state (as well as between
the Reno and Las Vegas campuses). The 56k line to San Diego
is expected to be operational about 15 March.
Also, during the month we expect start carrying IP traffic
over the 56k DECnet link to SAIC. This is a temporary measure
to provide a continuation of Internet access to SAIC after the
17 March removal of their ARPAnet host connection. It will be
replaced by a CERFnet link in early May.
Finally, in March we expect to convert our 56K SDSCnet link to
the Research Institutes of Scripps Clinic to a p4200 link
carrying both TCP/IP & DECnet.
by Paul Love (loveep@sds.sdsc.edu)
SESQUINET
No report received.
SURANET
SURAnet continues to increase in the number of sites connected
and in the number of networks advertised to the NSFnet.
At present there are 57 sites online and 71 networks are being
advertised to the NSFnet.
The current list of sites and networks can be obtained via
anonymous FTP from noc.sura.net, password guest, cd pub. File
name is "online".
by Jack Hahn (hahn@umd5.umd.edu)
Westine [Page 23]
Internet Monthly Report February 1989
WESTNET
1. T-1 Line Between University of Colorado at Boulder and
NCAR:
For large packets, we were experiencing high error rates
on the T-1 line between UCB and NCAR. We had originally
thought there was a problem with the Dowty Information
System T-1 CSU's, as we had reported last month. This
was incorrect; the problem was eventually tracked down to
problems in the US West span between UCB and the Table
Mesa Office. US West has provided a different span, and
the performance is now satisfactory.
2. New Networks:
The following networks were added to Westnet-East
USGS 192.41.213 U.S. Geological Survey (Denver)
VLA 192.43.204 Very Large Array Telescope (NM)
AFWL 129.238 Air Force Weapons Lab (NM)
3. Collection of Statistics with SNMP
Carol ward has this month begun using SNMP to collect
statistics in "production mode."
4. New 56 Kbps Circuit from University of Wyoming to
Colorado State University
The link was upgraded from a voice grade circuit using
compression modems (yielding 19.2 Kbps) to a 56 Kbps
subchannel on a microwave circuit. The microwave circuit
was a joint venture between UW and CSU for passing video
traffic.
by Pat Burns (pburns@super.org), and Carol Ward
(cward@spot.colorado.edu)