[IMR] IMR89-03.TXT MARCH 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). TABLE OF CONTENTS APPLICATIONS - USER INTERFACE . . . . . . . . . . . . . . page 3 AUTONOMOUS NETWORKS . . . . . . . . . . . . . . . . . . . page 3 END-TO-END SERVICES . . . . . . . . . . . . . . . . . . . page 3 INTERNET ARCHITECTURE . . . . . . . . . . . . . . . . . . page 3 INTERNET ENGINEEERING . . . . . . . . . . . . . . . . . . page 5 INTERNET MANAGEMENT . . . . . . . . . . . . . . . . . . . page 8 PRIVACY . . . . . . . . . . . . . . . . . . . . . . . . . page 8 SCIENTIFIC REQUIREMENT . . . . . . . . . . . . . . . . . page 8 DSAB . . . . . . . . . . . . . . . . . . . . . . . . . . page 8 BOLT BERANEK AND NEWMAN, INC., . . . . . . . . . . . . . page 9 CORNELL . . . . . . . . . . . . . . . . . . . . . . . . . page 11 ISI . . . . . . . . . . . . . . . . . . . . . . . . . . . page 11 MIT . . . . . . . . . . . . . . . . . . . . . . . . . . . page 14 MITRE . . . . . . . . . . . . . . . . . . . . . . . . . . page 14 Westine [Page 1]  Internet Monthly Report March 1989 NTA . . . . . . . . . . . . . . . . . . . . . . . . . . . page 14 NYSERNET . . . . . . . . . . . . . . . . . . . . . . . . page 15 SRI . . . . . . . . . . . . . . . . . . . . . . . . . . . page 16 UCL . . . . . . . . . . . . . . . . . . . . . . . . . . . page 16 UDEL. . . . . . . . . . . . . . . . . . . . . . . . . . . page 16 NSF NETWORKING. . . . . . . . . . . . . . . . . . . . . . page 18 UCAR/BOLT BERANEK and NEWMAN, INC., NNSC . . . . . . . . page 18 NSFNET BACKBONE (Merit) . . . . . . . . . . . . . . . . . page 18 NSFNET BACKBONE SITES & MID-LEVEL NETWORK SITES: BARRNET . . . . . . . . . . . . . . . . . . . . . . . . . page 19 CERFNET . . . . . . . . . . . . . . . . . . . . . . . . . page 19 CICNET. . . . . . . . . . . . . . . . . . . . . . . . . . page 19 UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN/NCSANET . . . page 22 JVNCNET, NORTH EAST RESEARCH REGIONAL NETWORK . . . . . . page 22 MERIT/UMNET . . . . . . . . . . . . . . . . . . . . . . . page 22 MRNET. . . . . . . . . . . . . . . . . . . . . . . . . . page 22 NATIONAL CENTER FOR ATMOSPHERIC RESEARCH AND UINIVERSITY SATELLITE NETWORK PROJECT . . . . . . . . . . . . . . . page 23 NORTHWESTNET. . . . . . . . . . . . . . . . . . . . . . . page 23 OARNET. . . . . . . . . . . . . . . . . . . . . . . . . . page 24 PITTSBURGH SUPERCOMPUTER CENTER . . . . . . . . . . . . . page 24 SAN DIEGO SUPERCOMPUTER CENTER . . . . . . . . . . . . . page 24 SESQUINET . . . . . . . . . . . . . . . . . . . . . . . . page 25 SURANET . . . . . . . . . . . . . . . . . . . . . . . . . page 25 WESTNET . . . . . . . . . . . . . . . . . . . . . . . . . page 25 Westine [Page 2]  Internet Monthly Report March 1989 TASK FORCE REPORTS ------------------ APPLICATIONS -- USER INTERFACE No report received. AUTONOMOUS NETWORKS We are proceeding with action items identified at our February meeting. Notes from that meeting are available from estrin@oberon.usc.edu. Deborah Estrin (Estrin@OBERON.USC.EDU) END-TO-END SERVICES No internet-related progress to report. Bob Braden (Braden@ISI.EDU) INTERNET ARCHITECTURE The following announcement has been sent to selected engineering and architecture mailing lists and is to appear in the March issue of the ACM Computer Communication Review. Workshop on the Future of the Internet System Architecture and TCP/IP Protocols 1-2 June 1989, University of Delaware Sponsored by the Internet Architecture Task Force and the Internet Activities Board The Internet Activities Board (IAB) has been guiding and coordinating the research and development activities of the DARPA/NSF Internet System for several years. The Internet Architecture Task Force (INARC) of the IAB has been asked to explore the inherent limitations in the existing Internet architecture and supporting IP/TCP protocol suite and how the lessons learned can be applied to future systems. The INARC will hold a two-day workshop on 1-2 June 1989 at the University of Delaware to explore these and related issues. While the emphasis of the workshop will be on the past and future evolution of the Internet system, specific issues relevant to other architectures, protocol suites and migration strategies may be discussed as well. Westine [Page 3]  Internet Monthly Report March 1989 Interested persons from all walks of network life are invited to attend. Participants will be encouraged to present short briefings on specific technical issues, including those suggested below, but this is not a requirement for admission. While some participants may be invited on the basis of their known expertise, biases and past vocalizations on these issues, participants outside the IAB, INARC and their dependencies are actively encouraged. In order to manage the local arrangements it is necessary that participants register their intent to attend by contacting the INARC chair: David L. Mills Electrical Engineering Department University of Delaware Newark, DE 19716 (302) 451-8247 mills@udel.edu.edu Areas of interest include, but are not limited to, the following: 1. Are the Internet architecture and protocols suitable for use on very high-speed networks operating in the 1000 Mbps range and up? If the network-level or transport-level protocols are not usable directly, can they be modified or new ones developed to operate effectively at these speeds? 2. Are the Internet addressing and gateway-routing algorithms adequate for very large networks with millions of subscribers? If not, is it possible to extend the addressing scope and/or develop new routing paradigms without starting over from scratch? 3. Can the Internet model of stateless networks and stateful hosts be evolved to include sophisticated algorithms for flow management, congestion control and effective use of multiple, prioritized paths? Can this be done without abandoning the estimated 60,000 hosts and 700 networks now gatewayed in the system? 4. Can the existing Internet of about 300 routing domains be evolved to support the policy and engineering mechanisms for many thousands of domains including education, research, commercial and government interests? Can this be Westine [Page 4]  Internet Monthly Report March 1989 done with existing decentralized management styles and funding sources? If not, what changes are needed and how can they be supported, given practical limits on infrastructure funding? Dave Mills (Mills@HUEY.UDEL.EDU) INTERNET ENGINEERING 1) The IETF will meet April 11-14, 1989 in Cocoa Beach, Florida. The meeting will be hosted by Kennedy Space Center. An abbreviated version of the agenda is below. 2) Responding to a vote of attendees at the Jan IETF meeting in Austin, the April meeting is 3.5 days. The first two days are devoted fully to Working Group sessions, the third day is devoted fully to technical presentations and network reports, and the concluding half day is devoted to reports from the Working Groups. 3) Highlights of Working Group status: - There are now 20 Working Groups. - A total of 15 Working Groups will meet in Cocoa Beach. - There are three new WGs meeting for the first time in Cocoa Beach. They are Network Management Services Interface, NOC Tools, and Host Dynamic Configuration. - Three other WGs have progressed to the point of making detailed presentations during the technical sessions on Thursday. These groups are Interconnectivity, Open Routing, and OSFPIGP. 4) Overview of WG status: IETF Working Group Status (April 1989) Working Groups RFC or Met Current Meeting Chair or POC Draft? Jan 89? Report? Apr 89? (address) ------------------------------------------------------------------------- Authentication Yes No No Yes Jeff Schiller (MIT) jis@athena.mit.edu CMIP-over-TCP (CMOT) Yes Yes Yes No Lee LaBarre (MITRE) cel@mitre-bedford.arpa DNS (new) Yes Yes Yes Yes Paul Mockapetris (ISI) Westine [Page 5]  Internet Monthly Report March 1989 pvm@isi.edu Host Requirements Yes Yes Yes No Bob Braden (ISI) braden@isi.edu Host Dyn. Config. (new) - - - Yes Ralph Droms (Bucknell) droms@sol.bucknell.edu Interconnectivity Yes Yes Yes Yes Guy Almes (Rice) almes@rice.edu Internet MIB No Yes Yes No Craig Partridge (BBN) craig@nnsc.nsf.net LAN Mgr MIB (new) No - Yes Yes Amatzia_Ben-Artzi@ spd.3com.com NM Ser. Interface (new) - - - Yes Jeff Case (UTK) case@utkcs2.cs.utk.edu NOC Tools (new) - - - Yes Bob Enger (Contel) enger@sccgate.scc.com NSFnet/Reg Monitoring Yes Yes Yes Yes Susan Hares (Merit) skh@merit.edu Open SPF-based IGP Yes No Yes Yes Mike Petry (UMD) petry@trantor.umd.edu Open Systems Routing Yes No No No Marianne Lepp (BBN) mlepp@bbn.com OSI Interoperability No Yes Yes Yes Ross Callon (DEC) callon@erlang.dec.com PDN Routing Group No No Yes Yes C-H Rokitansky roki@isi.edu Performance and CC Yes Yes Yes Yes Allison Mankin (MITRE) mankin@gateway.mitre.org Pt-Pt Protocol Yes Yes Yes Yes Drew Perkins (CMU) ddp#@andrew.cmu.edu ST and CO-IP No Yes Yes Yes Claudio Topolcic (BBN) topolcic@bbn.com TELNET Linemode Yes Yes Yes No Dave Borman (Cray) dab@cray.com User Services (New) No Yes Yes Yes Karen Bowers (NRI) bowers@sccgate.scc.com 5) Agenda for the April 11-14 IETF Meeting TUESDAY, APRIL 11th -- Working Group Sessions o OSPFIGP (Petry, UMD and Moy, Proteon) o Network Management Services Interface (Case, UTK and McCloghrie, TWG) o OSI Interoperation (Callon, DEC and Hagens, UWisc) o Performance and Congestion Control, TCP Subgroup (Mankin, Mitre) o Point-Point Protocol (Perkins, CMU and Hobby, UCDavis) Westine [Page 6]  Internet Monthly Report March 1989 o User Services (Bowers, NRI) o Authentication (Schiller, MIT and Rochlis, MIT) o LANMAN MIB (Ben-Artzi, 3Com) o Domain Name System WG (convened by Drew Perkins, CMU) WEDNESDAY, APRIL 12th -- Working Group Sessions o NOC Tools (Enger, Contel and Stine, Sparta) o Joint Interconnectivity and Open Routing WGs (Almes, Rice and Lepp, BBN) o Public Data Network Routing (Rokitanski, FERN) o Performance and Congestion Control (Mankin, Mitre) o ST and Connection IP (Topolcic, BBN) o Host Dynamic Configuration (Droms, Bucknell, P.Gross, NRI) o Joint Monitoring Access for (NSFNET) Adjacent Networks (Hares, Merit) THURSDAY, APRIL 13th o Everything You Ever Wanted to Know about OSPFIGP (including how to pronounce it) (Moy, Proteon) o An Interim Routing Architecture (or what to do until Open Routing gets here...) (Almes, Rice) o The Open Routing Architecture (Lepp, BBN) o Internet Report (BBN) o Growth of the Internet (DCA B600) o Report on the NASA Science Internet (Medin, Ames) o Report on the DOE Energy Science Network (Hain, LLNL) o NSFNET Report - Architectural Changes to NSFNET (Gerlich, MERIT) - Nifty NSFNET Stats, using NNStat (Hares, MERIT) o Arpanet Evaporation Timetable and An Overview of FRICC Initiatives (eg, NNT, RIB, and RIG) (P.Gross, NRI) o The DCA TCP/IP Certification Program (M.Gross, DCA-DCEC) o IP Header Compression (Van Jacobson, LBL) FRIDAY, APRIL 14th o Working Group Reports and Discussion o Concluding Plenary Remarks and Group Discussion o Tour of Kennedy Space Center 6) For more information about the IETF, please send to ietf-request@isi.edu. Phill Gross, IETF Chair (gross@sccgate.scc.com) Corporation for National Research Initiatives Westine [Page 7]  Internet Monthly Report March 1989 INTERNET MANAGEMENT No report received. PRIVACY No report received. SCIENTIFIC REQUIREMENTS No report received. DSAB ---- No report received. Westine [Page 8]  Internet Monthly Report March 1989 BOLT BERANEK AND NEWMAN INC. ---------------------------- WIDEBAND NETWORK Wideband Network activities this month consisted largely of the continued support of network operations and performance under conditions of heavy usage. In addition, reconfiguration of the network associated with the coming transition from a satellite- based to a terrestrial fiber-based transmission medium has begun. A prototype interface between the Sun Workstation based Wideband monitoring system and DARPA's Automated Network Monitoring (ANM) system was developed this month, thereby providing currently- available ANM graphical network management capabilities to the Wideband Network. The interface is implemented via an intermediate process which provides a translation service between the monitoring data format used by an ANM management module and the format used by the existing Wideband Satellite Network Poller (SNP). This connection allows ANM processes such as the Integrated Monitoring Display Process to access the management module for status and performance information from the Wideband Network. SATNET This month, SATNET performance was excellent. Statistics collected by ISI showed an average of 99+% uptime for the remaining two SIMPs. During this period, there was a hardware problem with the DCEC gateway, but the CSS gateway provided uninterrupted connectivity. On March 20, the Goonhilly SIMP was decommissioned by British Telecomm. The connections between the Butterfly gateway at RSRE and the gateways at BBN, NTA, and UCL continued to be stable. At this point, only the U.S. and Italian SIMPs are still operating. The new point-to-point link between CNUCE, Italy and DARPA, U.S. is expected to be operational in mid spring at which time these SIMPs will be retired. TERRESTRIAL WIDEBAND NETWORK The hardware and software development for the transition of the Wideband Network to terrestrial links continued to make progress this month. The installation work will begin the first week of April and continue as tail circuits between the backbone and the gateways become available. An initial Terrestrial Wideband Network will be run in parallel with the satellite-based Wideband with the two networks connected by a gateway at BBN. This arrangement will Westine [Page 9]  Internet Monthly Report March 1989 help to maintain connectivity between sites and minimize disruptions as we transition gateways to the new network. The Terrestrial Wideband Network will use one of several cross- country T1 trunks that are being provided as part of the National Networking Testbed (NNT). This trunk will go from Boston to San Francisco in an in-line sequence of hops. Wideband Packet- Switching nodes (WPS's) will be placed at the NNT Points of Presence (POPs) and in a few cases at user sites. T1 tail circuits will be used to connect the original set of Wideband sites (minus NOSC, DCEC, Lincoln Lab, JPL, Mitre/ESD, and MIT) to this backbone. The software for the Terrestrial Wideband Network will be fielded in two phases. The first phase will support most current Wideband services except for dynamic groups and reserved bandwidth streams. It will nonetheless support static groups and multimedia conferences. A second phase will follow the first and will implement possible features such as higher bandwidth trunks, e.g., 6 Mbits/sec DS2 trunks, and additional services for resource management. INTERNET R&D Martha Steenstrup has written a white paper with arguments on why congestion control is necessary and an outline of a congestion control algorithm with both feed-forward and feed-back controls. We are in the process of buying/acquiring simulator tools to begin testing the algorithm. The Open Routing Working Group (of the IETF) is now chaired by Marianne Lepp. The group has resolved the main issues of the top level architecture. The main unresolved issue is how to do data reduction. In February ORWG met jointly with the ANTF (Autonomous Systems Task Force) and got some excellent feed-back and guidance on policy issues from ANTF. During their own meeting time, they worked out a list of key issues to be decided in the protocol. Many of the questions have been answered, but it was useful to get together and discuss what is left. ORWG met again by teleconference this March. The topic of discussion was how to deal with size. A number of abstractions were discussed, but the issue has not been resolved yet. ORWG is preparing a functional specification which should be available by next month. Bob Hinden (Hinden@BBN.COM) Westine [Page 10]  Internet Monthly Report March 1989 CORNELL UNIVERSITY ------------------ A quiet month. We are putting together a suggested plan for what should be done with the RIGs in terms of initial policy-based routing. Work on Gated has been mainly testing and consolidating last month's EGP work. The main exception was major cleanup and enhancement of the NYSERNet Inc. SNMP interface to gated. Gated now generates SNMP traps through NYSERNet's snmpd for link up, link down and egp neighbor lost conditions. Jeff Honig and Scott Brim (jch@devvax.tn.cornell.edu,swb@chumley.tn.cornell.edu) ISI --- INTERNET CONCEPTS PROJECT Results indicate that, while the IP level congestion control algorithm is very effective at controlling congestion and is reasonably efficient in using available bandwidth, it unfairly distributes that bandwidth among competitive sources. Currently, a model gateway creates an SQ message for each datagram that causes a gateway queue overflow. The unfair distribution was not expected. The probability of receiving a SQ message from a congested gateway seems out of proportion to that source's use of the congested path. A question arose: Would fairness be markedly improved if instead, when an overflow occurred, a gateway randomly chose from its overflowing queue the message to discard and for which it builds a Source Quench? The source datagram that causes an overflow will not necessarily be discarded and provoke an SQ message in that case. The probablility of a source's receiving an SQ message is related to the recent reception history of the overflowing queue. A large series of simulations was run that seems to demonstrate a marked improvement in distribution of resource. A comparison of the distribution of process finishing times shows a much narrower spread when the random SQ approach is used. Greg Finn Jon Postel attended the CERFNET meeting at SDSC in San Diego, and the Calinet meeting in San Ramon, CA, 7-9, March. Jon Postel participated in the Collaboratory Workshop of NSF, in New York City, 12-14 March. Jon Postel gave a talk on "Los Nettos", at the Third IEEE Metropolitan Area Nets Workshop, Dana Point, CA, 28-30 March. Westine [Page 11]  Internet Monthly Report March 1989 Two RFCs were published this month. RFC 1094: Sun Microsystems, Inc., "NFS: Network File System Protocol Specification", March 1989. RFC 1096: Marcy, G., "Telnet X Display Location Option", Carnegie Mellon University, March 1989. Ann Westine (Westine.ISI.EDU) LOS NETTOS JPL was added as a new member of Los Nettos. The installation went without a problem. A T1 line for Rand is on order and the current delivery schedule is May 1. TRW is joining Los Nettos. With Pacific Bells new intra-LATA tariff, it is cheaper to run a longer T1 line from TRW (in GTE territory) to USC (in Pacific Bells territory) than having both local loops in GTE territory. NOSC will be joining Los Nettos as well. A line is on order. IBMs installation is delayed until their move to a new building. The move is currently anticipated in July. UCLA is now using Los Nettos as their only path to the Arpanet. Any sites wishing information about joining Los Nettos should contact Walt Prue at Prue@isi.edu or via phone at 213-822-1511. Walter Prue (Prue@ISI.EDU) MULTIMEDIA CONFERENCING PROJECT The multimedia teleconference facility in Washington was moved from the DARPA building at 1400 Wilson up the street two blocks to 1555. The new room is much larger and has a much better setup than the temporary installation at 1400. In conjunction with the move, a Widcom video codec was set up. We are adapting the packet video software to test the Widcom codec as an alternative to the Image30 codec to demonstrate improved video resolution. We have been helping BBN to test the latest version of their MMConf conferencing software. We have also worked on integration of MMConf and MMCC, the control software for voice and video conferencing. The new MMConf can present both plain text and multimedia Diamond/Slate documents. This should make it easier for teleconferencing users to prepare materials. Also, we joined the ranks of Slate multimedia mail sites, having coerced mmdeliver, the program responsible for multimedia mail delivery, to run on our Vax where mailboxes are stored. Westine [Page 12]  Internet Monthly Report March 1989 A new mailbox has been set up for people to request use of the teleconference facilities. It is VIDEO-CONF-REQUESTS@BBN.COM (or @ISI.EDU which will forward to BBN). Dave Walden, Eve Schooler, Steve Casner (djwalden@ISI.EDU, schooler@ISI.EDU, casner@ISI.EDU) NSFNET PROJECT Annette DeSchon fixed a number of minor BFTP bugs and released a new tar-file, version 1.12 or BFTP.12.tar.Z in the "pub/" directory on venera.isi.edu. She started work on the Multiple-host Background File Transfer Program, or "mbftp", for the Multimedia Conferencing Project. Bob Braden attended a two-day meeting convened by the FRICC to rewrite the NRN (National Research Network) program plan. This meeting was held in Annapolis and included representatives of the FRICC, the IAB, congressional staff, and the major common carrier companies. Braden chaired a one-day video teleconference of the End-to-End Task Force, using facilities donated by MCI, and also attended a one-day video teleconference of the IETF Open Routing Working Group, using the Internet packet video system. He also worked on arrangements for the upcoming IAB teleconference. Finally, we returned to work on the Host Requirements RFC, honing the text and making changes suggested by Mike Karels, Bill Barns, Charlie Lynn, Dave Borman, Paul Mockapetris, Jon Postel, and others. The version available for anonymous FTP from venera.isi.edu has changed almost daily. This will be the last report from the ISI NSFnet project. Bob Braden and Annette DeSchon (Braden@ISI.EDU, DeSchon@ISI.EDU) Westine [Page 13]  Internet Monthly Report March 1989 FAST PARTS FAST is a DARPA funded research project to develop an automated broker for the acquisition of standard electronic components. The FAST broker currently serves 50 project at 19 Universities, 4 companies, and 3 government agencies. FAST interacts with its users on the ARPANET via formated electronic mail messages. FAST continues to develop communication links with its suppliers using both X12 and commerical E-mail routes via MCI. We have recently begun to explore the automatic generation of FAXs from the SUN workstation. Paula Caruso presented "FAST PARTS" at the Electronic Contracting Conference of the National Contract Management Association, in Philadelphia, April 3, 1989. Alan Katz continued work on a Remote Execution Protocol and checked out release 3 of X windows, version 11 which was recently installed here. Alan attended the SCAE conference on Concurrent Engineering: CALS Phase II and Beyond on March 15-16 in Costa Mesa, California. Paula Caruso and Alan Katz (Caruso@ISI.EDU, Katz@ISI.EDU) MIT-LCS ------- No internet-related activity to report this month. Lixia Zhang (Lixia@XX.LCS.MIT.EDU) MITRE Corporation ----------------- No report received. NTA-RE and NDRE --------------- No report received. Westine [Page 14]  Internet Monthly Report March 1989 NYSERNET -------- An open seminar was held in Troy MAR29-31 on how to ->USE<- SNMP to manage a network. SNMP/MIB/SMI theory was covered in 4 hours and then the practical aspects of network management including several hours of "hands on" with representative agents (cisco,proteon,unix) and NMS's was completed in over 265 slides. The Draft of the pseudo public NYSERNet network operations manual was distributed as a supplement to the seminar. RFC 1089 was published on running SNMP over the Ethernet MAC layer. Work on a common SNMP Application Protocol Interface (API) by the SNMP authors is drawing to a close, posting to the snmp mailing list will be done before the Florida IETF. The API will make NMS tools highly portable across protocol implementations, mix and match tools should be available following RFC publication. THE TCP/IP NETWORK: - adding 12 more sites in the next 60 days - P4100's are now being used for all 9.6kbps and 56kbps leaf connections, one difficulty is their lack of support of the Proteon Serial line MIB error info which we've been using for the last 2 years on the p4200's. Very difficult to do without. - the root domain server C.NYSER.NET was moved to a dedicated SUN3/50, aside from the 10's of thousands of TCP connections from IBM's FAL/DNS implementation it is is as stable as named ever is. - we're finally responding to the pressure of PC/MAC users who are internet'ed, newer NYSERNet sites are primarily PC/MAC sites, we will begin operating a POP server and a PCMail Repository in the near future. "NYS" ARPANET: Canadian DND research network cutover from Rochester Butterfly/Arpanet connection to Proteon/NYSERNet in Rochester. Another circuit terminated in Buffalo is on order for the Canadian DND. Cornell, Columbia, UofR-CS, Rockefeller, all removed from Arpanet, outsanding for removal are: NYSER, GE-SYR, Hazeltine, Bellcore, RADC, BellLabs. All except for RADC are scheduled to be removed by 1May89. NYSERNet's ARPANet circuit is tested to Cambridge, awaiting PSN installation to our Gateway in Troy,NY. GE-SYR and Hazeltine will be cutover to NYSERNet. Westine [Page 15]  Internet Monthly Report March 1989 NNT: - both the Arpanet (long-haul) and the Wideband are moving to the NNT, most of this will be completed VERY early April. - Much wideband equipment (ButterFlies) will be located in the POP's for enhanced reliability and lower cost - DOE Internet is next Martin Schoffstall (schoff@rebel.nyser.net) SRI --- No internet progress to report this month. Zaw-Sing Su (Zsu@KL.SRI.COM UCL --- Jon Crowcroft attended and gave a paper on UCL work at a workshop on Telematique a Grande Vitesse (TGV) in Liege, Belgium. A number of Supercomputer Users throughout Europe are sponsering this series of workshops on high speed networking driven very much by the user requirements (e.g., CERN etc). John Crowcroft (jon@CS.UCL.AC.UK) UNIVERSITY OF DELAWARE ---------------------- 1. Much effort this month was spent on proposals and document preparation of various kinds. Our grads continue hacking theses and various odd jobs. Jeff Simpson is finishing up a Masters Thesis on a formal model and grammar for a policy- based routing protocol. Paul Schragger built a token-ring simulation environment in OPNET as a class project. Dave Mills attended the FRICC retreat in Annapolis, MD, and the IAB telemeeting in Washington, DC. 2. In extended discussions with Mike Little of SAIC we are working on a basic strawman architecture and engineering model for policy definition and implementation. The motivation for this study is to explore engineering methodologies appropriate for the policy requirements being developed by the ANTF. The strawman draws from the autonomous-confederation model of RFC-975 plus recent discussions in the ORWG and ANTF. One somewhat radical feature of the strawman is that the IP Westine [Page 16]  Internet Monthly Report March 1989 address space is reused in each confederation and routing between confederations is based only on the space of confederation numbers. 3. Activity continued this month testing the new NTP Version 2 beta implementations for Fuzzball and Unix systems. Various subtle bugs were found and corrected in the implementations and specification. Careful measurements made at several sites over periods of days to weeks have demonstrated that the new version has far superior stability and accuracy relative to the older versions. The present factors limiting accuracy appear to be instabilities in the crystal oscillators used in the time servers and in the various propagation modes used by the NIST broadcast time services. 4. Precision crystal oscillators have been ordered for two of our time-server hosts. These should improve frequency accuracy to the order of 10**-10 and, in particular, allow further investigation into the factors limiting stability and accuracy. Dave Mills (Mills@UDEL.EDU) Westine [Page 17]  Internet Monthly Report March 1989 NSF NETWORKING -------------- NSF NETWORKING UCAR/BOLT BERANEK AND NEWMAN INC., NNSC The NNSC is updating the contact information for local/campus NSFNET liaisons. Please send any changes to nnsc@nnsc.nsf.net. Karen Roubicek attended a meeting at NSF to plan a networking conference for end-users in April. by Karen Roubicek (roubicek@nnsc.nsf.net) NSFNET BACKBONE Merit/NSFNET Internetworking Seminar An Internetworking Seminar will be held April 17 and 18 at the Radisson Suite Resort at Hilton Head, South Carolina. This seminar is designed for campus liaisons and for those who help end users. Limited openings still exist. For further information or to register, please telephone 1-800- 66-MERIT or send electronic mail to 'nsfnet-info@merit.edu'. NSFNET Advanced Topics Seminar The NSFNET Internet Engineering Group sponsored a seminar in Ann Arbor on March 8 and 9. Eighty-four attendees participated in discussions of current networking issues. Presentors from across the country discussed such subjects as IS-IS routing, SNMP in the NSFNET, network monitoring tools, NSFNET architectural changes, and a number of other topics. In conjunction with the seminar, a working-group meeting of the Joint Monitoring Access for Adjacent Networks (JO-MAAN) was held and attended by some 50 people. This group discussed ways in which the NSFNET backbone, regional networks, and campus networks could cooperate to solve user problems expediently. Network Mapping Capabilities Enhanced The Internet Engineering Group is currently expanding the use of digital cartographical software in order to generate both interactive and hardcopy network maps. Software and data have been acquired from several sources and we can now produce maps of the NSFNET backbone as well as several regional networks. The Westine [Page 18]  Internet Monthly Report March 1989 interactive version is based on the X Window System and will be used for network planning and other purposes. We are collecting geographic coordinates from regional networks in order to produce these maps and welcome this information from the regionals. Please send comments or inquiries via electronic mail to 'ie@merit.edu' or telephone 1-800-66-MERIT. Updates to NSFNET-IS Remote Query Database Two enhancements to the Merit-NSFNET Information Services remote query database have been made. The CONTACTS command has been expanded and the TOPOLOGY command has been added. Complete documentation may be obtained by sending electronic mail to either: nis-info@nis.nsf.net or nis-info@merit (Bitnet) and specifying "HELP" as the first line of the message. (The "subject" field is ignored.) If assistance is needed in using the server, send electronic mail to either: userhelp@nis.nsf.net or userhelp@merit (Bitnet) Comparison of Packet Counts ------------------------------------------------------------ February 1989 / March 1989 Packets In Packets Out February 573,390,288 593,669,705 March 745,599,984 755,073,841 % increase 30% 27% ------------------------------------------------------------ by Patricia G. Smith (PGS@nis.nsf.net) NSFNET BACKBONE SITES & MID-LEVEL NETWORK SITES BARRNET No report received. CERFNET Technically Speaking During the month of March, CERFnet placed equipment orders for Westine [Page 19]  Internet Monthly Report March 1989 circuits, cisco gateways and CSU/DSUs. CERFnet expects to receive notice from vendors regarding firm delivery dates shortly. The installation of the backbone has been rescheduled for the first week of May. Installations for remaining sites will begin the last week of May and first week of June. A 56 kilobit link was initiated from the San Diego Supercomputer Center to Qualcomm, Incorporated, a San Diego based company, on March 28th. The cisco AGS at the California State University Chancellor's Office SWRL facility and CSU Fullerton are running the 7.1 release. They have found a major (instant, consistent crash) with DECnet on the 7.1 release. DECnet circuits are not automatically figured at this time. However, Appletalk routing is now being done by the cisco. Other Network Activity On March 1st and 2nd a CERFnet representative attended the FARNet meeting held at Stanford University. On March 7th CERFnet held a plenary meeting at the San Diego Supercomputer Center. Items discussed included the Fiscal Year 1989 Budget, the Los Nettos/CERFnet Cooperation Plan, the CERFnet Topology, and the installation schedule. The next meeting is scheduled for July 10th, 1989. A signed agreement was received from the California Institute of Technology. CERFnet accepted an Associate Membership from Custom Product Design, an Irvine based company. University of California at Irvine, will manage their membership. by Karen Armstrong (armstrongk@sds.sdsc.edu) CICNET The Committee on Institutional Cooperation's NSF-supported CICNet became fully operational on 3 March 1989. CICNet is comprised of T1 circuits, provided by MCI, and Cisco routers on campuses at the University of Chicago, the Urbana-Champaign and Chicago campuses of the University of Illinois, and the remaining Big Ten University campuses, with the exception of Purdue University. Last December, Merit assumed the responsibility for the implementation of CICNet, with the assistance of the CICNet Westine [Page 20]  Internet Monthly Report March 1989 Technical Committee, composed of members from each of the participating universities. Jessica Yu, of Merit's Internet Engineering staff was the coordinator and the principal person at Merit overseeing the engineering aspects of the implementation. Cathy Aronson, formerly a network operator in Merit's NOC, began her duties as the person in our Internet Engineering group with principal responsibilities for CICNet in January of this year, and is taking over Jessica's CICNet role in the engineering aspects of this project. On 1 March 1989, Joel Maloff assumed duties as the new Executive Director of CICNet, replacing Barbara Wolfe, who has moved on to a position at Southern Methodist University. Joel was formerly associated with the National Telecommunications Network, a consortium of fiber-optic-based communications providers. Two CICNet positions within Merit's Information Services will soon be filled, and three new operators in Merit's Network Operations Center have been hired to accommodate our expanded operations and management responsibilities. The full eleven-node network has been up and running for four weeks and the Merit NOC is closely monitoring CICNet. During this period there were no major outages on the network. Some data regarding pre- and post-CICNet connectivity situations, collected by Jessica Yu, give an indication of the improvement provided by the new network: Ping Delay Times: Ann Arbor to Ohio State hosts OSU Hosts round trip delays (min/ave/max) to 35.1.1.30 pre-CICNet CICNet ------------------------------------------------------ 128.146.1.4 84/85/88 16/16/22 128.146.8.60 94/118/337 24/29/39 128.146.8.62 91/97/110 21/29/28 130.101.2.2 131/136/141 65/87/121 Merit's CICNet staff can be reached at CICNet@merit.edu, or at 1- 800-66-MERIT. A current PostScript version of the CICNet topology map (CICNet.PS) is stored on the MAPS directory on the nis.nsf.net (35.1.1.48) Information Services host machine, and is obtained via anonymous FTP. current PostScript version of the CICNet topology map (CICNet.PS) is stored on the MAPS directory on the nis.nsf.net (35.1.1.48) Information Services host machine, and is obtained via anonymous FTP. by James C. Sweeton (SWEETON@merit) Westine [Page 21]  Internet Monthly Report March 1989 UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN/NCSANET No report received. JVNCNET NORTHEAST RESEARCH REGIONAL NETWORK No report received. MERIT/UMNET No report received. MIDNET No report received. MRNET Bethel College was formally accepted as a member of MRNet this month. Congratulations to Bethel College. ETA Systems has upgraded their link to MRNet to use Cisco routers. This will enable more of the ETA internal network to access MRNet and the Internet. ETA also expects their traffic to increase as more people are able to easily access MRNet. The NSF grant for equipment to connect Carleton College and St. Olaf College was approved. These schools are moving towards upgrading their connections. Several ad hoc committees were formed during the March general meeting. These committees include: o A business plan committee which will explore alternative organizational structures for MRNet as well as recommend the range of services MRNet ought to offer to its members. o A membership committee which will, is some sense, examine what the mission of MRNet ought to be and by implication who our members are likely to be. o A bylaws committee which will review the MRNet bylaws and recommend any updates which are believed warranted. by Tim Salo (tjs@msc.umn.edu) Westine [Page 22]  Internet Monthly Report March 1989 NATIONAL CENTER FOR ATMOSPHERIC RESEARCH AND THE UNIVERSITY SATELLITE NETWORK PROJECT NCAR recently acquired a new Cray XMP-18 and is bringing up the UNIX operating system UNICOS on it. This allows the possibility of Internet access via TCP/IP. Since the LAN for the supercomputers was historically a NSC Hyperchannel, practical access to the Internet is only possible via a TCP/IP Hyperchannel <-> ethernet NSC router known as the EN641. Unfortunately, the NCAR local class B net, is a bridged network. If an attempt is made to use the same class B net for the Hyperchannel and ethernet side, then many of the "subnets" on the ethernet side are unreachable because the EN641 insists on a subnet mask as the only way to direct packets to the proper interface. The EN641 does not have the ability to use a static table of {subnet, interface} pairs to direct packets. As a result, NCAR will have to acquire a class C address for hosts on the Hyperchannel. by Don Morris (morris@windom.ucar.edu) NORTHWESTNET During March, NWnet experienced one significant (leased line related) outage, but otherwise performed very reliably. That outage made us realize that although we had designed the network for redundancy, this redundancy existed only for TCP/IP; our DECnet design was not robust in the presence of single link failures since several areas spanned multiple sites. We have renumbered, but the moral for multiple- protocol networks seems to be that one must be careful to analyze each protocol topology separately. Router upgrades to Proteon's 8.1 release are in progress, and should be completed during April. After analysis of traffic patterns and network reliability, we have concluded that the present ring topology (circumference 9) of NWnet does not serve our needs: it puts several sites 2 or more analog- modem hops away from the NSS (implying very high RTTs to NSFnet) without contributing substantially to network uptime. We anticipate shifting to a star topology later this spring. Our analysis continues. One promising possibility is the use of dialup IP to create fallback circuits when failures partition the network. Our network manager, Boeing Computer Services, is finally beginning to provide us with some performance and traffic data on the network, which should make further analysis substantially easier. by JQ Johnson (jqj@hogg.cc.uoregon.edu) Westine [Page 23]  Internet Monthly Report March 1989 OARNET No report received. PITTSBURGH SUPERCOMPUTING CENTER PSC is now providing NSFnet to ARPAnet connectivity via a new gateway, psc-gw4 (192.5.146.4, 10.7.0.14). Psc-gw4 carries all of the traffic from NSF only networks to ARPA/MIL only networks and about half of the traffic from ARPA/MIL only networks to NSF only networks (The rest is carried by uxc.cso.uiuc.edu). Psc-gw4 has been very stable and has not had problems running out of resources. It and its 56k link to the ARPAnet have been relatively unstressed. However our connection to the ARPAnet, PSN14, is seriously congested. Psc-gw4 discards 20% of the traffic presented from the NSFnet because PSN14 refuses the traffic. This is the long term average: During prime time it often discards more than it delivers. The 56k link almost never reaches 50% utilization. All PSCnet east Proteon P4200 have been upgraded to version 8.1. PSCnet west Proteons are running version 8.0. We will be upgrading them to 8.1 when we resolve DECNET configurations. by Matt Mathis (mathis@fornax.ece.cmu.edu) SAN DIEGO SUPERCOMPUTER CENTER The 56k link between UNV and SDSC has been installed and tested. The UNV gateway cisco is now scheduled to arrive in early April. As planned, during the month we started carrying IP traffic over the 56k DECnet link to SAIC (net 192.5.8). This is a temporary measure and will be replaced by a CERFnet link in early May. Our 56K SDSCnet link to the Research Institutes of Scripps Clinic has been converted from SDSCnet to a p4200 link. It is currently carrying IP traffic. DECnet will be added in the near future. RISC is net 192.42.82. A Class B address (132.249 - SDSClan) has been assigned to the San Diego Supercomputer Center. We will be converting from out existing Class C address (192.12.207) but a Flag Day has not yet been selected. Paul Love San Diego Supercomputer Center loveep@sds.sdsc.edu Westine [Page 24]  Internet Monthly Report March 1989 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 58 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) WESTNET 1. Carol Ward attended the SNMP workshop at NYSERNet, and David Wood attended the FARNet meeting at Stanford. Both meetings were well received by the attendees. 2. The T-1 connection between the University of Colorado and NCAR (finally) seems to be reliable and free of significant error. 3. We had a couple of cisco CPU cards fail near the end of this month. This is worrisome to us; we are wondering if such occurrences are endemic in the hardware. We would appreciate hearing from any of you who have experienced statistically significant events with cisco hardware -- especially multiple failures within a short duration. 4. The microwave link between the University of Wyoming and Colorado State University (which replaced an analog circuit) is working very well, yielding much improved performance. by Pat Burns (pburns@super.org), David C. M. Wood (dcmwood@spot.colorado.edu)