[IMR] IMR88-03.TXT MARCH 1988 INTERNET MONTHLY REPORTS ------------------------ The purpose of these reports is to communicate to the Internet Research Group the accomplishments, milestones reached, or problems discovered by the participating organizations. This report is for research use only, and is not for public distribution. Each organization is expected to submit a 1/2 page report on the first business day of the month describing the previous month's activities. These reports should be submitted via network mail to Ann Westine (Westine@ISI.EDU) or Karen Roubicek (Roubicek@NNSC.NSF.NET). BBN LABORATORIES AND BBN COMMUNICATIONS CORPORATION --------------------------------------------------- WIDEBAND NETWORK A Wideband Butterfly Gateway was installed at Lincoln Laboratory early in the month. The new gateway, connecting the ARPANET to the Wideband Network via the Lincoln Laboratory BSAT, will provide backup ARPANET connectivity for the Cronus DOS cluster at ESD/Mitre. A coast-to-coast meeting of the Distributed Systems Advisory Board was supported by the Wideband Network and the multimedia conferencing facilities at BBN and ISI on March 25. This was the first such multimedia conference to employ commercial color video codecs and echo cancellation units. A new Wideband Network test site was brought up on the satellite channel via shared access to BBN's Wideband earth station. This site includes a BSAT running recently developed software that implements an interface to SIMNET, DARPA's real-time interactive military vehicle simulation program. This SIMNET-BSAT interface will support experimentation in and demonstrations of the application of Wideband packet satellite technology to the connection of geographically dispersed local area networks of SIMNET simulators. The problems that had been encountered in the new interface between the DCEC Butterfly Gateway and the DCEC BSAT have been resolved. The gateway is now providing stable SATNET-Wideband connectivity. Westine [Page 1]  Internet Monthly Report March 1988 INTERNET R&D We started working on adding support for Internet Multicast to the Butterfly Gateway. This has some interesting design issues regarding how to do multicast over networks which do not support multicast services and how to maintain and distribute the multicast routing database. The internet passed the 400 network mark this month. It is currently about 420 networks. Earlier in the month we installed a patch in the LSI-11 gateways to increase the routing tables from 400 to 500. VAX NETWORKING David Waitzman has started work with Steve Deering of Stanford on mechanisms to support internet-wide multicast groups. SATNET The SATNET has been very stable through the month of March. We have had no unscheduled outages of the SATNET SIMP's or PSP terminal hardware. The availability of the SATNET was again above 99% from tests run by ISI. A new SIMP configuration tape was sent to Goonhilly to prepare for the upgrade of the Goonhilly to RSRE line to 64 Kb Kilostream service. We are waiting for BT to notify us that the line is in place. The UCL to RSRE line was upgraded to the new service and operated for a short time. A problem developed with the line and it was turned back over to BT. Bob Hinden (Hinden@BBN.COM) ISI --- Internet Concepts Project Jon Postel participated in the FRICC meeting on Networks in San Diego, March 3, 1988. Jon Postel attended the IAB meeting in San Francisco, March 20-21. Jon Postel attended the Surfnet meeting at UC Irvine, March 7. Paul Mockapetris gave a presentation at the Energy Research Network Workshop in Oakland, CA., March 9. Paul Mockapetris and Joyce Reynolds attended the IETF meeting at the San Diego Super Computer Computing Center, March 1-2. Westine [Page 2]  Internet Monthly Report March 1988 Two RFCs were published this month. RFC 1049: Sirbu, M., "A CONTENT-TYPE HEADER FIELD FOR INTERNET MESSAGES", CMU, March 1988. RFC 1051: Prindeville, P., "A Standard for the Transmission of IP Datagramsand ARP Packets over ARCNET Networks", McGill University, March 1988. Ann Westine Multimedia Conferencing Project Packet video is up and running with the Image 30 video codec boards from Concept Communications. We have modified the on- board firmware of the Image 30 to implement the HDLC serial line protocol required to interface to the Butterfly. Substantial changes have also been made in the packet video program (PVP) that runs in the Butterfly to accommodate the packetization scheme for the new codec. The Image 30 boards give us several advantages compared to the ISI-built experimental codecs we used previously. First, they produce color video instead of black and white. Second, we have four of them so we will be able to expand soon from two to four teleconference sites. Third, the Image 30 takes two camera inputs, so we have at each site the room-view camera plus a copy-stand camera. To control camera switching and full-screen/quadrant selection, we have implemented a video control panel using software buttons in a small window sharing the Sun screen with MMCONF/Diamond. The new codecs were used for a tele-meeting of the Distributed Systems Architecture Board on March 25. The resolution of the video image wasn't bad for the room view, but was insufficient for viewgraphs on the copy stand. We will be implementing a high-resolution, low-frame-rate mode for this purpose. Steve Casner went to Boston to instalal packet video equipment at BBN, March 20-21. Steve Casner, Eve Schooler, Dave Walden (Casner@ISI.EDU) Brian Hung continues to work on the echo canceller algorithm. Brian did some simulation studies on the effect of arithmetic errors on the stability of the algorithm. The studies showed that there were two places in his algorithm where arithmetic errors were significant enough to cause instability. In one place it involved accumulation of 16 bit sums where 32 bit Westine [Page 3]  Internet Monthly Report March 1988 accumulation was called for and the other place involved truncation errors where round-off was required. Brian is currently implementing the PCM/binary conversion routine for the echo canceller algorithm. Brian Hung (Hung@ISI.EDU) NSFNET Project Bob Braden and Annette DeSchon produced Release 2.1 of NNStat, which includes the complete source. It also contains an extensively-revised document which should be more helpful to users of the package. A separate document on the internal algorithms of statspy was also written; both are being published as ISI reports. Annette also made some improvements in the Background File Transfer program BFTP. Braden attended the IETF meeting in San Diego, Feb 29 to Mar 3, to join the Working Group on Host Requirements. He also attended a one-day video teleconference of the IAB. Bob Braden and Annette DeSchon (Braden@ISI.EDU, DeSchon@ISI.EDU) Supercomputer and Workstation Communication Project Alan Katz worked on a Mandlebrot set graphics viewer that runs under the X Window system. Alan Katz (Katz@ISI.EDU) MIT-LCS ------- No relevant progress to report for March. Lixia Zhang (Lixia@XX.LCS.MIT.EDU) NTA-RE and NDRE --------------- 1. Paal Spilling received a message from the central part of the administration that COMSAT now has the equipment available and that the repointing of the antenna at Tanum will be effec- tuated a week or two from now! The installation of the 9.6 kb/s line to RSRE has been delayed and will definitely not be Westine [Page 4]  Internet Monthly Report March 1988 in place when the antenna is being repointed. NTA is also looking into the possibility of setting up a 64 kb/s line via Tanum - Etam/Roaring Creek or a Ku-band connection directly from Norway to Washington. The latter possibility will not be available before June. It is also not clear how the US part of such a line should connect into Internet. This will be taken up with DARPA/BBN. 2. One part of the Norwegian research internet, covering the universities and some research institutions, are operational. 10 - 15 Cisco boxes are utilized as IP-routers. 3. NTA-RE is in the final planning stage for a project in packet-switched satellite communication. A three-site testbed will be established hopefully at the end of this year, using WBNET technology, to interconnect local area networks at the three test sites. After a test and experimentation period of two to three years, it is the meaning to put the network into public service. Paal Spilling (paal@tor.nta.no) SRI --- Internet Research A paper titled "An Open-Systems Model for Computer-Supported Collaboration", by J.J. Garcia-Luna-Aceves, Earl J. Craighill, and Ruth Lang, was published in the Proceedings of the 2nd IEEE Conference on Workstations. This paper explores the functions and structure of systems for computer-supported collaborative work (CSCW) to share information and to use its resources in an integrated manner for their mutual collaboration. It describes MOSAIC, a model for CSCW systems; and discuss a prototype designed to support collaborative decision making. J.J. Garcia-Luna-Aceves (garcia@tsca.istc.sri.com) Distributed Applications We have installed the TACTICS II distributed system (first reported in July and August 1987 issues of Internet Monthly) at the Combined Arms Center (CAC) Fort Leavenworth, KC and at Com- munications Experiment Command (CECOM) Fort Monmouth, NJ. The CAC system includes a Sun 3/280 fileserver, a Sun 3/60 fileserver and nine diskless Sun 3/60 client workstations on an Westine [Page 5]  Internet Monthly Report March 1988 Ethernet LAN. The TACTICS II software installed on these sys- tems consists of five independent databases each sharing a sin- gle copy of the resource management system and name server. These five databases were configured to support four classroom accounts and a software demonstration account. For each data- base, a database server is running that allows remote access and updates a directory that contains tactical information. The CECOM system includes a Sun 3/260 fileserver, two Sun 3/50 clients, and a Sun 2/120 fileserver across an Ethernet. Similar to the CAC system configuration, the TACTICS II software here consists of a single copy of the resource management system shared by two fileservers and two clients. It is different from the CAC configuration in that a single database was created for the entire LAN but is also replicated across the two fileservers. Client applications generating queries to this database would select the database server "closest" to it based on a best server heuristic and host performance monitoring scheme. Ron Lee (rlee@spam.istc.sri.com) UCL --- Operations: The UK triangle is now operating on 64kbps on all lines and performance and availability are good. We are now running tests on a subnet with a microvax and a Cisco gateway with an Ethernet and X.25 line. This is the system that will be moving to ULCC to run the UK JANET to US NSF internet relay. Research: The work on TCP/IP and SATNET measurements is now fairly complete, and a report will be forthcoming. The Thorn directory service is now being heavily used by the Department for a number of yellow page type services, and is being evaluated for supporting ISO network management information bases. John Crowcroft (jon@CS.UCL.AC.UK) Westine [Page 6]  Internet Monthly Report March 1988 UNIVERSITY OF DELAWARE ---------------------- 1. In an effort to evaluate the effect of the source-quench policy recently implemented for the NSFNET Backbone fuzzballs, Mike Minnich set up an experiment involving a Sun workstation with the modified Jacobsen/Karels TCP and a UDel fuzzball connected between the Sun Ethernet and a 9600-bps upstream serial line. The experiment showed a thirty-percent reduction in queue size for the upstream line when source quench was activated. While confirming the policy does work, the experiment also showed the need for further experiments to evaluate the effectiveness for various types of traffic and traffic distributions. 2. Another experiment was designed to evaluate the current performance of the Backbone with respect to previous performance, as well as gauge the effectiveness of recent changes in queueing disciplines, including the source-quench implementation. In July 1988 the Backbone aggregate traffic load per queue ranged from 0.3 to 4.0 packets per second, with a mean of 2.0. At that time the preemption rate was .06 and timeout rate .03 percent. Now, several months after quench was installed, the load ranges from 2.6 to 9.2 packets per second, with a mean of 4.7. The preemption rate is 0.37, timeout rate 0.11 and quench rate 0.27 percent. The traffic on the sixteen 56-Kbps internodal trunks has doubled from 31.5 to over 71 packets per second, but the aggregate loss rate is still well below the system objective of one percent. A memo describing the experiment and results was circulated to the various engineering lists. 3. A third experiment was designed to reveal the cause of reported glitches in Backbone connectivity, but here definitive results remain tantalizingly elusive. As shown above, performance between the Backbone fuzzballs themselves is generally very good, although questions remain about apparent lockup of the DEQNA interfaces at the Ethernet demarcations. However, it was confirmed with ICMP echo and NTP time messages that at least some of the NSFNET/ARPANET gateways occasionally experience delay spasms of 30 seconds or more, presumably due to known X.25 problems, but gateway logs do not readily confirm this as the cause. Occasionally severe churn (rate of change of routes and/or distances) is observed (via EGP) in the BBN core gateways for some NSFNET clients, but not others, even though both are reported by the same gateway. These issues remain to be resolved. Westine [Page 7]  Internet Monthly Report March 1988 4. The Network Time Protocol (NTP) implementation for the fuzzball was again revised in light of recent experience with broken clocks (at one point only one out of five radio clocks was working), severe mains-frequency jitter and wander and just plain bugs. In an effort to find out why the mains- frequency clocks were so much worse when synchronized with NTP than when synchronized by Hellospeak, a simple phase-lock loop simulator was implemented using the existing NTP loop filter. The results showed the loop was in fact unstable with the US and German national power grids, so the loop filter was redesigned. Stable, updated fuzzware is now in place at most primary and secondary servers. 5. The document "Network Time Protocol (Version 1): Specification and Implementation" has again been revised in light of several suggestions and the above changes. Mike Petry has completed a companion NTP daemon for Unix 4.3bsd, which is now in test. 6. Dave Mills attended the Gigabit Working Group meeting at Stanford, the IAB telemeeting at MCI Washington, an NRC meeting on survivable telephone networks at NAS Washington and presented a tutorial on Internet protocols at INFOCOM 88 in New Orleans. A paper on the seventeen-year swampy trek of the Fuzzball was completed and submitted to SIGCOM 88. Dave Mills (Mills@UDEL.EDU) Westine [Page 8]  Internet Monthly Report March 1988 NSF NETWORKING -------------- UCAR/BBN LABS NNSC The NNSC published the third issue of the NSF Network News and announced a two-day tutorial on the domain system to be held at the NNSC on April 28 and 29. Craig Partridge attended the adhoc Network Management Review meeting, the Internet Engineering Task Force meeting and gave a presentation at the Interface '88 Conference in Chicago. By Karen Roubicek (roubicek@nnsc.nsf.net) NEW NSFNET BACKBONE As of 24 March, MCI has installed and completed initial data circuit testing of all T1 links to the 13 NSFNET sites ahead of their 1 April scheduled completion date. Traffic is flowing on the test network among Ann Arbor; Reston, VA; Milford, CT; and Yorktown, NY. An early version of the NSS software is currently being installed on the four nodes of the test network as well as the first operational node currently being build for Ann Arbor. The Ann Arbor test network node currently uses only four RT/PCs: a Routing Control Processor (RCP), an External Packet Switching Processor (E-PSP), and two Packet Switching Processors (PSPs) with synchronous cards. The NSS is currently located next to the IBM 4381 Information Services (IS) machine, which is also connected to the network. Testing of FTP on the IS machine is nearly completed. Information should soon be providable via anonymous FTP and SMTP mail from the IS machine. The NOC IBM 4381 should also be installed shortly. Westine [Page 9]  Internet Monthly Report March 1988 NSFNET Backbone and Test Network T1 circuits: * Yorktown (IBM) / \ / \ Seattle / \ * Milford (IBM) * Test- * Reston (MCI) \ \ nwk / \ \ / * Ithaca \ Ann Arbor\ / / \ \ /---------------------*---\ / \ \ / \ / \ \ /Boulder \ / \ \ \ *-------*-------* Pittsburgh * ------* Princeton \ SL City / Lincoln /| / \ / / | / * \ / /-----------*--------/ | * Palo \ \ / / Urbana-Ch. | College Park Alto \ \ / / / \ \/ / / \------*------/ / San Diego / * Houston In a move designed to simplify final installation at the remote sites and optimize the path to network production, most of the hardware and software for all NSS nodes will be assembled, undergo basic testing, and be re-packaged for final shipment at Merit facilities in Ann Arbor. A combined team of IBM and Merit people has begun assembling the Ann Arbor pilot node. Starting 11 April we plan on three nodes per week to be assembled, tested, and packaged for shipment to sites. The assembly will continue until all thirteen nodes are completed. The IDNX and Verilink CSUs will be shipped directly to the remote locations for installation and end-to-end physical network testing by MCI and IBM prior to the shipment of NSS equipment. By Laura Kelleher (lkelleher@merit.edu) Westine [Page 10]  Internet Monthly Report March 1988 NSFNET BACKBONE SITES CORNELL UNIVERSITY THEORY CENTER As of 27 March, 1988, there were 157 networks being advertised on NSFNET. The networks that were added in February and March are listed following this report. Jeff Honig and Scott Brim are working hard with the IBM and Merit folks to modify the gated software in support of the new NSFnet backbone. This includes support for the use of gated on the gateways to the regional networks and the portions of gated used in the Nodal Switching Subsystems. In the past two months, development work on gated has included many EGP fixes, enhancement of EGP tracing, additional validation of networks sent and received via EGP, martian filtering for all protocols, fixes and improvements to the interface timeout code and improvements to the debugging code. A number of these modifications have recently been made available to several sites for beta test. More testing is in order before another version is released. IBM and Cornell have continued joint investigation on 8232 performance and TCP loops in VM/TCP and VM/XA/SP1/TCP. Mike Hojnowski gave a presentation at the Spring SHARE meeting on VM TCP/IP User Experiences. Doug Elias is preparing the Statspy data processing code and documentation for distribution to interested parties. The video tapes of the Networking Workshop, held in August 1988, with Vint Cerf, Bob Braden, and Dave Mills are ready for distribution. Alison Brown, Associate Director of Advanced Computing and Networks, has accepted the position of Associate Director of the Ohio Supercomputing Center. She will be acting as Director of OARnet (Ohio Academic Resources Network) in her new role. She will assume her new position in mid-April. With her departure, those who normally communicate with Alison on networking topics should now contact Scott Brim. Westine [Page 11]  Internet Monthly Report March 1988 Networks added to the NSFNET Routing Tables during February and March. BARRNet 36 Stanford 128.32 UC-BERKELEY 128.114 UC-SANTACRUZ 128.120 UC-DAVIS 128.218 UC-SAN-FRAN NORTHEASTNet 128.197 BU 129.10 NORTHEASTERN 129.133 WESLEYAN 128.36 YALE 192.26.88 YALE NORTHWESTNet 128.208 UWASH 129.95 OGCNET NYSERNet 129.3 OSWEGO 192.31.254 ALFRED SURAnet 192.31.192 IDA/SRC 129.28 ETA 129.43 NCIFCRF 129.57 CEBAF 128.239 WM-NET WESTNet 128.123 NMSU 129.24 UMN 129.121 NMT 129.138 NMIMT MISCnets 128.200 UC-IRVINE 129.176 MAYO 192.35.81 WISC-PARK By Martyne Hallgren (martyne@tcgould.tn.cornell.edu) Westine [Page 12]  Internet Monthly Report March 1988 UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN/NCSANET The 56Kbps line to Univ of Wisconsin-Milwaukee was installed March 31. We estimate that they will be on-line approximately April 12, when their Proteon router arrives. The T1 for the University of Chicago should also be up and running within the next week or two, with Northwestern's T1 not far behind. Fermi Nat'l Lab is installing a 14.2Kbps IP connection (no DECnet on this one) and we expect this to be up shortly. UIUC ARPANET connection went into production. Now gateways NSFNET and ARPANET for all points west of Champaign, without own connections. Corrected long standing configuration problem with Fuzzbal DMV11's dip switch wrong cause long recovery after error. These two items have significantly improved NSFNET performance. By Ed Krol (krol@uxc.cso.uiuc.edu) JOHN VON NEUMANN NATIONAL SUPERCOMPUTER CENTER No report received. NATIONAL CENTER FOR ATMOSPHERIC RESEARCH AND UNIVERSITY SATELLITE NETWORK PROJECT The Naval Research Laboratory (NRL) has finally come on-line as a USAN site. The directly connected network at NRL is 128.60 and the USAN gateway node is 128.116.30.1. By Don Morris (morris@scdsw1.ucar.edu) PITTSBURGH SUPERCOMPUTING CENTER IP and Decnet routers are now in place between Carnegie-Mellon University's campus net and PSCnet. Arrangements are being made to install a Proteon P4200 between the University of Pittsburgh net and PSCnet. During March a new kernel with a beta version of the ACC 5250 driver was brought up on PSC-GW in preparation for a new version doubling the number of available virtual circuits. This new driver along with resolution of the PSN 7.0 and new End-to-End code on the ARPANET should allow for better and more consistent connectivity. PSCnet connections using Proteon P4200's running version 7.4 are now in place on the campuses of seven of our Academic Affiliates. We are providing name service for those institutions that have requested it. Westine [Page 13]  Internet Monthly Report March 1988 Efforts are being made toward connecting PSC-GW2, our second Arpanet gateway, to PSN #21 which recently became functional during February on our premises. Two trunks are currently installed and connected, and a third has been installed but has not yet been connected to the PSN here. Four DS/1 circuits have been terminated here for the new NSFnet backbone connections. By David O'Leary (oleary@godot.psc.edu) SAN DIEGO SUPERCOMPUTER CENTER A group from "Task Force 100" will visit SDSC on April 1 to activate our PSN's trunk connections. This is much sooner than last month's Tiger Team had indicated so, to whoever this should go, thanks for the speedup! We are now supporting NorthWestNet via our trusty p4200. We have upgraded the software to 7.4b. No anomallies were observed during the month. NWNet is scheduled to be turned-up in GATED on April 1. We have ported client RSH to the Cray OS, CTSS. It is in test this month and should be in production use sometime during April. By Paul Love (loveep@sds.sdsc.edu) NSFNET MID-LEVEL NETWORKS BARRNET (No report received) JVNCNET (Refer to JVNNSC backbone report) MERIT/UMNET An Ethernet IP subnetwork was added on a Wayne State University Secondary Communications Processor (SCP) (WSC1). At Oakland University an Ethernet link for IP support will be added to the OK Primary Communications Processor (PCP). We are now using a Cisco gateway for our HDH ARPANET link as well as for the Ethernet USAN connection. By Laura Kelleher (lkelleher@merit.edu) Westine [Page 14]  Internet Monthly Report March 1988 MIDNET Not much new has happened. I am now using the SGMP software provided by Jeff Case and Ken Key at Tennessee.It is proving usefull in monitoring the routers in MIDnet. Several sites have gotten their name servers running and are looking foreward to the tutorials on Bind at the end of April. The individual sites are making more progress all the time in getting their local networks extended throughtout the campuses. By Dale Finkelson (dmf@fergvax.unl.edu) MRNET MRNet activities during March consisted of reviewing membership applications for charter members and discussing facilities requirements (security, hardware, etc.). Members currently online, who thus have direct acces to NSFnet include ETA Systems, the Mayo Foundation, the Minnesota Supercomputer Center, and the University of Minnesota. Cray Research is also indirectly connected to MRNet at this time, and also has access to NSFnet. Cray is in the process of upgrading their connection to MRNet to a more direct one. Connections for Control Data Corporation, 3M, Carleton College, and St. Olaf College are pending. MRNet's representatives attended the presentation in Ann Arbor of the Merit/IBM/MCI high speed NSFnet trunk network, now under construction. They also participated in the business activities of the Federation of Research Networks, and applied to be full members of same. A host somewhere on NSFnet, sending routing-information packets claiming to come from an MRNet IP address, had been causing very unreliable access from MRNet to much of the NSF Internet, possibly for several months. The errant host has not been located, but we are working around the problem. NSFnet no longer appears to be unusable. Ken Carlson (kcg@uf.msc.umn.edu) NCSANET (Refer to UIUC & NCSAnet backbone report) NORTHWESTNET (No report received) Westine [Page 15]  Internet Monthly Report March 1988 NYSERNET NYSERNet had the following topology: Clarkson Syracuse--+ | NISC | | | Rochester--------Cornell---------RPI---Albany | || || | ....Alfred || || Buffalo...Fredonia || || | ....Oswego || || | || || Binghamton || +-------- || ------StonyBrook | || | || | || | || CUNY--NYTEL/NSMAC--Columbia======NYU==Rockefeller | |\ | | /| | || | | \ | NYNEX/S&T / | | || | | \ BNL / | | WP/CO | | \ / | | | | +-------------POLY---+ | | | | | | | +------------------------------+ | | | +--------------------------------------+ || ==== || T1 || \ | / 56kbits .... 9.6kbits In March: Mark Fedor participated in a "routing integration" meeting at IBM Yorktown Heights on regional network to NSS connectivity. About 70 licenses were distributed for our SGMP/SNMP implementation. Two new members joined: GE Corporate Research and Development and Grumman. By Martin Schoffstall (schoff@nisc.nyser.net) Westine [Page 16]  Internet Monthly Report March 1988 OARNET As February came to a close, OARnet (Ohio Academic Resources Network) was connected to the NSFNet backbone via a link to The Pittsburgh Supercomputer Center. OARnet is a regional network connecting over 20 Colleges and Universities in the state of Ohio with the Ohio Supercomputer Center via fiber optic lines. A variety of networking protocols are supported providing access to national and international networks. The supported protocols are: TCP/IP, Decnet, and NJE bisync (BITNET). OARnet has applied for membership in the NSFNet Federation. By Paul Buerger (paul@rigel.osc.ohio-state.edu) SDSCNET (Refer to SDSC backbone report) SESQUINET The complete initially proposed SesquiNet configuration, now augmented by Prairie View A&M, has been operational now for several months. The following campus networks are being served, and are advertised via EGP to the core: Baylor College of Medicine 128.249 BCM-Technologies 192.31.88 Houston Area Research Center 192.31.87 Prairie View A&M University 129.208 Rice University 128.42 Texas A&M University 128.194 Texas Southern University 192.31.101 and the University of Houston 129.7 The serial line from NSFnet/NCAR to SesquiNet/Rice has been operational for several months, and routes to SesquiNet via NSFnet are now being advertised. Performance is very good. One accomplishment this month has been an analysis of those destination networks advertised both by NSFnet and by the Core Gateways. In the case of each network reachable through both of these national networks, we have used Ping (very conservatively) to study the optimal route to take on average. The results are used to statically favor either NSFnet or the Arpanet on a network by network basis. We dynamically obtain reachability information (via EGP for the Arpanet and via Hello for the NSFnet); if a given network is not reachable via the statically favored route, the Westine [Page 17]  Internet Monthly Report March 1988 other is used. We are also testing cisco's support for 1822-HDH Arpanet support. We have found and had fixed a small number of bugs in this support, and are now using it operationally. We will soon resume our testing cisco's support for dual protocol (IP and DECnet) routing. By Guy Almes (almes@rice.edu) SURANET (No report received) WESTNET 1. This has been a month of great activity for Westnet schools, beginning with a meeting on March 8, of the Westnet Steering Committee, consisting of University Computer Center Directors from all sixteen Westnet sites. This was the first formal meeting of this group, and entailed establishing an adminis- trative structure, planning for the installation of hardware and circuits, and establishing the parameters under which Westnet will operate. A great deal of information transfer also took place, concerning the Backbone activity, the status of campus LAN's, and possible steady-state models for Westnet when NSF funding ceases. 2. The New Mexico universities and New Mexico Technet now have full access to the Internet, and the situation appears to have stabilized without much difficulty. 3. We are exploring means for the "western half" of Westnet to obtain access to the Internet between the time (in April) when the circuits are installed, and the time (in July) when the IBM NSS is scheduled to become operational at the University of Utah. Sergio Hecker has given permission for these "western half" schools to route to the Internet through the satellite link at the University of Arizona to JVNNC. However, we are still exploring the possibility of routing through the ARPANET IMP at the University of Utah. This latter path would be more desirable, as it would minimize reconfiguration of routing tables in July. Westine [Page 18]  Internet Monthly Report March 1988 4. cisco has asked us to be a beta test site for their version of Marty Schoffstall's SGMP, which is expected to be available to us in April. 5. The University of Colorado at Boulder and Digital Equipment Corporation jointly hosted a networking conference entitled "Net Show Spring '88." The conference was attended by over 100 people from diverse locations. By Pat Burns (pburns%csugreen.bitnet@cunyvm.cuny.edu) Westine [Page 19]  Internet Monthly Report March 1988 TASK FORCE REPORTS ------------------ APPLICATIONS -- USER INTERFACE The task force held its third meeting February 10-11 at Information Sciences Institute. The major topics of discussion were user interface architecture, voice integration, teleconferencing, and document interchange. The resulting highlights were: - There has been some convergence toward a reference model for user interface software, as documented in K.A. Lantz, et al. "Reference Models, Window Systems, and Concurrency," Computer Graphics, April 1987. - This convergence may, in fact, be mildly counter-productive, in that the model as presented in the referenced document does not adequately address issues surrounding collaborative work, in particular, sharing of applications in the context of a real-time teleconference. Fortunately, both SRI and Olivetti are addressing this problem. - Interest in the PC-based voice server has blossomed, with servers now in place at Olivetti, Sun, ANSA, and ISI---in addition to the Media Lab. Various PC-compatibles have been employed, with two different voice boards---the Dialog/2 board from Dialogic Corp. and the TI Speech Board. The only difference between servers based on the two speech boards is that the TI board does not support pause detection. The server library interface is in the process of being refined, including the specification of upload/download functions. An RFC is (still) forthcoming. - Olivetti Research Center, in collaboration with Chris Schmandt, is designing a version of the voice system that eliminates the PC. It will consist of an AT-bus voice board coupled with a UNIX server process. The server process is being designed in the mold of a window system---with multiple voice "channels", for example. Unfortunately, this has necessitated considerable redesign of the server interface. - Steve Casner (casner@isi.edu) has formed a working group to assess the state of the art in "workstation video". People actively engaged in the integration of live video into contemporary workstation environments should contact him as soon as possible. Westine [Page 20]  Internet Monthly Report March 1988 - It was agreed that the "integrative" approach to desktop teleconferencing promoted in K.A. Lantz, "An Experiment in Integrated Multimedia Conferencing," Proc. Conference on Computer-Supported Cooperative Work, December 1986, has been vindicated (although not necessarily to the exclusion of other approaches) by the many similar systems developed subsequently---including Rapport at AT&T Bell Labs, Shared X at HP Labs, VISEX at MCC, and the MOSAIC reference model at SRI International. However, many problems remain, including: maintenance of consistent context in the face of replicated applications; conference management; voice and video integration; and the integration of teleconferencing within the framework of a uniform model for collaborative system (see second point above). - There is need for additional work in the area of "media synchronization", that is, the synchronization of multiple data streams being transmitted on multiple communications channels (logical or physical). This is an issue whenever the communications channels differ in delay or bandwidth characteristics, or the sending and receiving hosts differ markedly in performance. Joe Sventek of the ANSA project is the most involved of the task force members in these issues. Complete minutes will be posted to UI-INTEREST momentarily. The next meeting will be June 20-23 in the Bay Area. Keith Lantz (Lantz@SCORE.STANFORD.EDU) AUTONOMOUS NETWORKS The Autonomous Networks Task force is planning a partial meeting in conjunction with the IEEE Symposium on Security and Privacy in Oakland California, April 18-20. We will hold a full meeting in California (probably LA) during the summer. Our current focus is on articulating IRI/FRICC requirements for accounting and access control. Deborah Estrin (Estrin@OBERON.USC.EDU) Westine [Page 21]  Internet Monthly Report March 1988 END-TO-END SERVICES No progress to report this month. Bob Braden (Braden@ISI.EDU) INTERNET ARCHITECTURE Activity was low this month, but sunspot numbers are up. Dave Mills (Mills@HUEY.UDEL.EDU) INTERNET ENGINEERING 1. Network Management Review Committee On February 29, 1988, an ad hoc review committee on network management met at the request of the IAB. Attendees consisted primarily of 1) representatives of the CMIS/CMIP, SGMP and HEMS network management efforts and 2) representatives of the IAB and IETF. The meeting was chaired by Vint Cerf. The purpose of the meeting was to review the various network management activities and to recommend to the IAB a course of action for both near-term and longer-term network management development. Vint Cerf has prepared a report summarizing the resulting conclusions and recommendations. This report has now been issued as RFC1052. Some of the major points, in brief, include: o designation of SGMP (possibly with some extensions) as the focus for near-term Internet/network management, o focusing on CMIS/CMIP as the longer-term solution, o tasking the IETF with the expeditious formation of two working groups: 1) to develop a Management Information Base (MIB) for the TCP/IP protocols to be used jointly by SGMP and CMIS/CMIP, and 2) to consider what extensions, if any, might be needed to SGMP to address any specific vendor concerns and to ease the eventual transition to CMIS/CMIP. In accordance with these recommendations, the IETF has begun the formation of the above two working groups. Craig Partridge has been asked to chair the MIB working group. For more information on that activity, contact Craig directly Westine [Page 22]  Internet Monthly Report March 1988 (craig@nnsc.nsf.net). The group to consider extensions to SGMP (now renamed to SNMP, for Simple Network Management Protocol) is still being formed. More detailed information on the SNMP extensions group will be announced as it develops. Since this work will depend in part on the MIB, it is appropriate for the MIB group to begin meeting first. 2. March 1-3 IETF Plenary On March 1-3, 1988, the IETF met at the San Diego Supercomputer Center. The meeting was hosted by Paul Love, who continued the trend of excellent local arrangements and fine weather set by earlier meetings. The agenda was distributed in last month's report. The Proceedings are still in preparation. 3. Previous Proceedings We have established a procedure by which IETF Proceedings will be available through the NIC. They currently have Proceedings from the November meeting at NCAR and the July meeting at MITRE. Contact Mary Stahl (stahl@sri-nic.arpa) for information on how to obtain copies. 4. New IDEAS There have been 10 new IDEAS issued since the March 1-3 meeting. IDEAS can be retrieved by anonymous FTP from the directory at SRI-NIC.ARPA. The file IDEA-INDEX-ABS.TXT contains an annotated index of all IDEAS. o IDEA009.TXT - `Exterior Gateway Protocol (EGP), Version 3', Marianne Gardner (BBN) and Mike Karels (UCBerkeley) o IDEA010.TXT - `A Laboratory For Testing DOD Protocol Implementations: It's Architecture And Methodologies', John Swanson and Jose Rodriguez (UNISYS McLean Research Center) o IDEA011.TXT - `A Simple Network Management Protocol', Jeffrey Case (U of TN), Mark Fedor (NYSERnet Inc.), Martin Schoffstall (RPI), and James Davin (Proteon Inc.) o IDEA012.TXT - 'Network Management for TCP/IP Network: An Overview', A. Ben-Artzi (Sytek) o IDEA013.TXT - 'Structure and Identification of Management Information for the Internet', Lee LaBarre (MITRE) o IDEA014.TXT - `Kerberos Authentication and Authorization Westine [Page 23]  Internet Monthly Report March 1988 System', S. P. Miller, B. C. Neuman, J.I. Schiller, and J.H. Saltzer (all of MIT) o IDEA015.TXT - `Host Extensions for IP Multicasting', S. E. Deering (Stanford) o IDEA016.TXT - `TELNET LOCALEDIT OPTION', Dave Borman (Cray Research) o IDEA017.TXT - `ISO Presentation Services on top of TCP/IP-based Internets', Marshall T Rose (The Wollongong Group) o IDEA018.TXT - `System Load', Inder Sidhu, N Arunkumar (Bridge) Phill Gross (gross@gateway.mitre.org) INTERNET MANAGEMENT No report received. PRIVACY As noted in the February report, the IAB Privacy Task Force had a productive two-day meeting at Lawrence Livermore National Laboratory in Livermore, California on 2 and 3 March. Attendees were: Dave Balenson, Curt Barker, Don Brinkley, Morrie Gasser, Steve Kent, John Linn, Dan Nessett, and Steve Wilbur. A number of messages were exchanged among the task force membership during March to continue discussion of topics raised at the meeting, particularly with regard to detailed RFC-1040 implementation and specification issues. Steve Kent travelled to SRI on 4 March and met with members of the IETF authentication working group (AWG) that afternoon. Mike St. Johns from the DDN PMO was present, along with Jeff Schiller and John Rochlis from Project Athena (MIT). Agreement was reached that key management should be handled independently from application of cryptography to specific protocols. Thus the AWG will work with other IETF members to develop specifications for securing SNMP (the recently renamed SGMP), e.g., specifying which fields are to be encrypted and or MACed and how IVs and key ids shall be represented in messages. The specs will indicate how to look up the appropriate key in a local key cache, based on addresses in the message or key ids. Issues associated with how the key cache is filled will NOT be addressed in the SNMP specifications. Thus one can begin with manual key distribution for initial testing (or for small systems) and migrate to various automated schemes as time passes. Thus adoption of Kerberos or other key management schemes as part Westine [Page 24]  Internet Monthly Report March 1988 of the SNMP spec is not considered appropriate. The next privacy task force meeting is scheduled for Wednesday-Thursday, 15-16 June 1988 at DEC, Littleton, MA. John Linn (Linn@CCY.BBN.COM) ROBUSTNESS AND SURVIVABILITY No report received. SCIENTIFIC COMPUTING No report received. SECURITY No report received. TACTICAL INTERNET No report received. TESTING AND EVALUATION No report received.