[IMR] IMR87-11.TXT Westine [Page 1]  ~ NOVEMBER 1987 INTERNET MONTHLY REPORTS ------------------------ The purpose of these reports is to communicate to the Internet Research Group the accomplishments, milestones reached, or problems discovered by the participating organizations. This report is for research use only, and is not for public distri- bution. Each organization is expected to submit a 1/2 page report on the first business day of the month describing the previous month's activities. These reports should be submitted via network mail to Ann Westine (Westine@ISI.EDU) or Karen Roubicek (Roubicek@SH.CS.NET). BBN LABORATORIES AND BBN COMMUNICATIONS CORPORATION --------------------------------------------------- WIDEBAND NETWORK BSAT Releases 6.1 and 6.2 were distributed to the Wideband Network sites this month. These releases primarily consisted of minor bug fixes. In addition, the operational BSAT software is now capable of supporting up to twelve Wideband sites. Work has started on the development of a faster and more reliable booting mechanism for the BSATs. This mechanism will exploit the Butterfly "RAMFile" capability distributed with Chrysalis Operating System Release 3.0. With the addition of non-volatile Multibus RAM Westine [Page 1]  Internet Monthly Report November 1987 hardware to the BSAT, the RAMFile facility can be used to cache BSAT application and Chrysalis operating system software and critical BSAT state through complete system reboots, including reboots resulting from site power outages. RAMFile-based booting is expected to reduce the required BSAT boot time from about 20 minutes to less than one minute. SATNET The SATNET continues to perform well. Roaring Creek, Tanum and Goonhilly had an availability of 99% from tests run at ISI. Fucino's availability was lower, 97%, due to a PSP outage early in the month which took the site off the air. The upgrade of the line between RSRE and UCL to 64 Kbps Kilostream service has been delayed. We are still waiting for the line to be installed between the two sites by BT. When a completion date for the installation of the line has been determined, we will reschedule the upgrade of the gateway hardware. INTERNET RESEARCH & DEVELOPMENT The EGP servers on the ARPANET were upgraded to DEC LSI 11/73's during November. The results have been wonderful. The ARPANET overall average round trip delay (for 1300-1700 EST on 11/24) as compared with a month earlier went from 1474msec to 442msec. The average round trip delays to the IMP's with the EGP servers went from 9033msec to 413msec. The total internode throughput went from 379 Kbps to 468 Kbps. For comparison, the 4-hour average is substantially greater than the 413 kbps total internode throughput for the peak hour in a 5- working-day CUMstats collection in August. At that time that was the largest throughput ever recorded for a single hour for the ARPANET. Another result of the upgrade was a dramatic reduction in the datagram drop rates of the ARPANET EGP servers and the DDN Mailbridges. Note that AERO, MINET, and YUMA, which are the MILNET EGP servers have not been upgraded. We hope to do this in December. AERO MINET YUMA BBN2 ISI PURDUE 11/12 6.56% 4.48% 6.63% 12.33% 14.25% 7.84% 11/13 0.90% 0.62% 5.06% 5.21% 12.81% 3.73% 11/14 0.40% 0.13% 2.70% 2.34% 7.32% 3.65% 11/15 0.25% 0.24% 2.61% 1.44% 5.95% 3.39% 11/16 1.14% 0.90% 5.68% 3.53% 10.80% 2.73% 11/17 1.50% 1.46% 8.96% 3.08% 9.32% 3.17% 11/18 1.12% 0.54% 11.85% 4.23% 9.85% 2.79% 11/19 1.78% 0.89% 11.01% 2.48% 5.97% 3.75% Westine [Page 2]  Internet Monthly Report November 1987 11/20 3.55% 1.13% 12.39% 2.29% 3.34% 4.95% 11/21 3.33% 0.42% 9.81% 0.75% 1.54% 0.92% 11/22 2.65% 0.68% 7.78% 1.33% 1.37% 2.17% 11/23 2.84% 0.86% 12.77% 1.42% 2.76% 2.93% 11/24 4.28% 1.21% 12.99% 1.74% 2.21% 2.10% 11/25 6.08% 1.39% 11.76% 1.56% 0.72% 1.45% 11/26 4.05% 0.36% 5.37% 0.70% 0.49% 0.72% MILARP MILBBN MILDCE MILISI MILLBL MILSAC MILSRI 11/12 20.04% 11.56% 7.89% 14.47% 3.01% 5.25% 6.50% 11/13 16.30% 11.22% 5.66% 22.20% 1.08% 5.10% 6.90% 11/14 12.80% 5.48% 2.34% 13.15% 7.36% 1.97% 1.45% 11/15 14.21% 4.46% 2.05% 9.00% 12.10% 2.37% 1.21% 11/16 23.58% 12.88% 9.08% 15.60% 1.67% 3.97% 4.92% 11/17 26.62% 11.59% 5.01% 12.34% 1.42% 6.09% 3.65% 11/18 28.24% 13.63% 7.04% 13.01% 1.97% 3.25% 4.10% 11/19 18.34% 8.23% 6.63% 6.33% 1.32% 1.71% 3.88% 11/20 4.97% 7.31% 9.50% 6.34% 0.27% 2.08% 6.03% 11/21 2.51% 1.74% 3.98% 0.68% 0.27% 1.00% 0.50% 11/22 1.78% 1.36% 1.91% 0.31% 0.20% 0.79% 0.40% 11/23 9.29% 4.57% 5.06% 1.02% 0.95% 3.23% 3.18% 11/24 8.94% 4.69% 2.99% 3.00% 0.58% 3.65% 1.44% 11/25 9.44% 3.25% 3.08% 2.57% 0.75% 4.08% 1.85% 11/26 4.99% 1.94% 2.99% 1.53% 0.94% 1.10% 0.57% Many thanks to Mike Petry at UMD who supplied two of the DEC LSI- 11/73. Robert Hinden (Hinden@PARK-STREET.BBN.COM) ISI --- Internet Concepts Project The Intermail mail forwarding system was moved from C.ISI.EDU to A.ISI.EDU. Although there were some interruptions in service due to hardware and software problems, Intermail appears to be running reliably once again. Westine [Page 3]  Internet Monthly Report November 1987 Seven RFCs were published this month. RFC 1028: Davin, J., J. Case, M. Fedor, and M. Schoffstall, "A SIMPLE GATEWAY MONITORING PROTOCOL", November 1987. RFC 1030: Lambert, M., "ON TESTING THE NETBLT PROTOCOL OVER DIVERS NETWORKS", November 1987. RFC 1031: Lazear, W., "MILNET NAME DOMAIN TRANSITION", November 1987. RFC 1032: Stahl, M., "DOMAIN ADMINISTRATORS GUIDE", November 1987. RFC 1033: Lotter, M., "DOMAIN ADMINISTRATORS OPERATIONS GUIDE", November 1987. RFC 1034: Mockapetris, P., "DOMAIN NAMES - CONCEPTS AND FACILITIES", November 1987. RFC 1035: Mockapetris, P., "DOMAIN NAMES - IMPLEMENTATION AND SPECIFICATION", November 1987. Ann Westine (Westine@ISI.EDU) Multimedia Conferencing Project The ISIC machine is fading from existence, so our "WBCTST" hourly probing of the Wideband Net and SATNET has been moved to ISIA. Since ISIA is not directly connected to the ARPANET, the extra gateway hop may decrease the probe success rate some. After ISI-GATEWAY was upgraded to an 11/73, though, any decrease seems slight. Steve Casner attended the Telecon VIII conference in Anaheim, CA, Nov. 10th, and saw the Concept Communications video codec in operation. Steve got positive answers to technical questions about interfacing this codec into our system for packet video. Steve is proceeding with the purchase of these codecs to replace the experimental codecs built at ISI for use in multimedia conferencing. This will allow us to expand from two packet video sites to four over the next several months. Steve Casner (Casner@ISI.EDU) Brian Hung is looking at implementing echo cancellors for the Multimedia Teleconferencing using the NEC PD7720 Signal Processing Interface. Right now he is reading a number of papers in order to come up with an algorithm to be implemented Westine [Page 4]  Internet Monthly Report November 1987 on the NEC PD7720 chip. Brian Hung (Hung@ISI.EDU) NSFNET Project Bob Braden and Annette DeSchon continued working on the statistics collection programs for the NSFNET. Bob Braden attended the Network Program Advisory Group meeting at NSF, 11-13 November. Annette DeSchon attended the IAB Autonomous Networks Task Force meeting at MIT in Boston, 4-6 November. Annette DeSchon and Bob Braden (DeSchon@ISI.EDU and Braden@ISI.EDU) Supercomputer and Workstation Communication Project Alan Katz wrote a few more utilities for GNU Emacs. Alan also has been investigating the possibility of making GNU Emacs a split-editor (with part of it running on a remote machine and part locally) in order to do performance experiments over the Wideband net. This would be helpful in determining how to best split a distributed editor and also to discover how tolerable using an editor over the Wideband net can be made. GNU Emacs might be very nice for this since it is very modular and is also very easy to customize. Alan has been investigating further issues in equations representation. Alan Katz (Katz@ISI.EDU) MIT-LCS ------- No report received. MITRE Corporation ----------------- No report received. NTA & NDRE ---------- No report received. Westine [Page 5]  Internet Monthly Report November 1987 SRI --- No report received. UCL --- Research: Work in the Admiral project is continuing using the formal specification languages Z and CSP, to design an N-Reliable Multicast protocol based on Birmans ?CAST primatives, together with a robust Group management scheme, based on Deering and Cheriton`s work at Stanford. Some initial performance figures for Remote Operations (ROS) over our full ISO stack (ISODE over X.25, or TP0/TCP) show it to be an order of magnitude slower than existing RPC systems for Request/Response type activity. Some of this overhead is for philospical acceptable reasons. We are investigating where the rest of the time is going. Infrastructure: The "UK Triangle" upgrade to 64kbps all round is in hand. UCL now have a second ethernet on our SATNET Butterfly, which has helped simplify our access control. We are now running Bind 4.7.3+ a bit for Domain Name Service, and should hopefully see better logical availability. John Crowcroft (jon@CS.UCL.AC.UK) UNIVERSITY OF DELAWARE ---------------------- 1. The massive disruptions of the NSFNET community previously reported have abated further during the month. Persistent observation, review of NSFNET Backbone logs and pursuit of zoo events turned up several specific areas in which stability was improved by further tuning of the Backbone fuzzballs. However, X.25 ARPANET access paths continue to suffer occasional disruptions due to congestion of virtual-circuit resources. 2. One of the Backfuzz improvements involved the mechanism used to measure baseline delays between nodes. Even though the Westine [Page 6]  Internet Monthly Report November 1987 mechanism is not traffic sensitive, it had become degraded due to link-level retransmissions caused by noisy circuits, humungus routing updates and massive overloads. The mechanism was modified to use pre-engineered delays on the most congested paths. Further improvements in stability may be possible upon close examination of the local routing environment at each Backbone site. 3. The "floating default" routing mechanism mentioned last month has now been implemented throughout the hellospeaking world of the Backbone and its dependencies. However, not all ARPANET gateways that could help drain the swamp are actually doing that, with the result that some paths continue to suffer severe congestion. 4. During the month several instances of fuzzbusters, defined here as bogon routing packets, occured where local routing packets (both RIPspeak and hellospeak) somehow escaped their local nets, wandered around the Internet and wound up on another local net. All fuzzballs, including the NSFNET Backbone machines, were modified to include fuzzbuster detectors (illegal in Virginia) which destroy these things on contact; however, the original points of escape have yet to be found. 5. Investigation and experiment on Internet congestion control continues. An analysis of the various logs showed a dramatic instance where a Big Blue machine (which presumably did not respond to ICMP Source Quench), drenched a Backbone machine, while a Craymonster (which did), didn't even get it wet. The incident suggests that the mechanisms prototyped in the fuzzball and some hosts really can produce worthwhile improvements in system utilization and service. 6. Development continues on the Dissimilar Gateway Protocol (DGP). Mike Little and Woody Woodburn have produced a draft specification, which is circulating now for review. The next task is testing the prototype in a controlled environment and later in the real Internet. 7. Dave Mills attended the NSF NPAG meeting in Washington. Planning continues for the INARC Workshop, to be held at BBN on 17-18 December. See the INARC report elsewhere for further information. Dave Mills (Mills@HUEY.UDEL) Westine [Page 7]  Internet Monthly Report November 1987 NSF NETWORKING -------------- NSF NETWORKING UCAR/BBN LABS NSF NETWORK SERVICE CENTER (NNSC) We completed and started distributing the second issue of the NSF Network News. To request additional copies, please send a message to . Craig Partridge attended the Internet Engineering Task Force meeting in Boulder and Karen Roubicek attended the meeting of the Federation of American Research Networks in Pittsburgh. Karen Roubicek (roubicek@nnsc.nsf.net) NSFNET BACKBONE SITES CORNELL UNIVERSITY THEORY CENTER No report received. UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN Received first loads of P4200 version 7.4 and it didn't load. There are two important reasons this version is of merit. One is the subnet routing problem. The other is that in looking at metric reconstruction as suggested by STRWG it appears virtually impossible without 7.4's split horizon. Line to Minnesota Research Network is due 5 Dec as well as our link to IMP 94. Connection of note is that MRN will also be closely connected to IMP94. Link to UIUC will be via P4200s. by Ed Krol (krol@uxc.cso.uiuc.edu) JOHN VON NEUMANN NATIONAL SUPERCOMPUTER CENTER This report is designed to inform the JVNC Consortium and JVNCnet network members as well as the Internet community of monthly status of the JVNCnet network. For a copy of the full report, send mail to "JVNCnet-nic@jvnca.csc.org". The gateways were available an average of 94.29 % of the time. There were many outages this month due to mostly hardware problems on the VAXs, T1 mux problems and routing problems with the UB routers, some of these required time to be fixed since some sites don't have 24 hours coverage. Still, the percentage of availability has been high, and overall performance of the network has been good. Westine [Page 8]  Internet Monthly Report November 1987 Due to congestion on the NSFNet backbone we diverted traffic between Penn State and NCAR via JVNCnet and University of Colorado (a backdoor path), allowing users at Penn State to access NCAR more reliably. Plans are on their way to bring up the new extension of the JVNCnet network in the North East. This extension includes: Yale University Wesleyan University, Dartmouth College, Boston University, Northeastern University, and University of Massachussets (Amherst). The schedule for bringing these six new sites is January 1988. These sites will connect at Harvard, Brown and JVNC making two double rings, and will be combination of T1 and 56kbps lines (being the T1 the existing circuits). INFORMATION: JVNCnet NOC: "net@jvnca.csc.org" (JVNCnet Network Operations Center) JVNCnet NIC: "JVNCnet-nic@jvnca.csc.org" (JVNCnet information) * CSC Consortium: Princeton University, MIT, Harvard, Brown, University of Pennsylvania, Rutgers, IAS, Columbia, University of Rochester, NYU, Penn State, University of Arizona, University of Colorado. ** NRAC (Newark Remote Access): New Jersey Institute of Technology (NJIT), Stevens Institute of Technology, University of Medicine and Dentistry of New Jersey (UMDNJ). by Sergio Heker (heker@jvnca.csc.org) NATIONAL CENTER FOR ATMOSPHERIC RESEARCH AND UNIVERSITY SATELLITE NETWORK PROJECT The connection of NRL (Naval Research Laboratory) in Washington, D.C. to the USAN Satellite network is complete. Most of the Bit Error Rate Tests are completed. The 56 Kbps land line, the cisco routers, and DSU/CSUs for the SESQUINET connection to USAN is in place. Hookup and testing is expected to be complete by December 9. The connection will be to an existing cisco router that also services the USAN connection to the University of Colorado. This connection will give SESQUINET access to NSFNET since an NSFNET node is connected to USAN. The addition of SESQUINET will bring to eleven the number of gateways on USAN. by Don Morris (morris@scdsw1.ucar.edu) Westine [Page 9]  Internet Monthly Report November 1987 PITTSBURGH SUPERCOMPUTING CENTER The PSC Fuzzball continued to perform well during November despite many changes to software. A concern was the "flapping routes" throughout the month, especially to the other backbone nets and other critical nets. Problems with virtual circuit limitations using ACC's X.25 interface to the PSN and BBN's move to a new software revision for the PSN's have caused numerous headaches on PSC-GW, but with reconfigurations of the net and advertising some NSF networks at other ARPANET gateways the situation has stabilized. New versions of software and firmware for the ACC board will be installed soon. These will not fix the virtual circuit limitation problem. New gated versions and configurations have been installed throughout the month. PSC-GW forwarded about 800,000 packets per day between the ARPANET and NSFnet. The installation of the satellite dish for PSN #21 here has been completed and tests are being run. We have had a line installed to the CMU PSN which we are willing to use pending action by DCA. We have received approval to connect another gateway machine to the PSN here. The new gateway machine has been built and is running the newest ACC code soon to be installed on PSC-GW. The T1 local loop connecting us to the Litel Point of Presence in Pittsburgh has been installed in preparation for connecting the western sites of PSCnet. We will install a Proteon gateway at Litel's Cleveland POP early in December and as soon as local loops are completed the rest of PSCnet will be installed. Another gateway machine has been built to provide better access for our CMU users to our center and the NSFnet. A gated configuration for this machine is being coordinated with the NSF NOC. by David O'Leary (oleary@morgul.psc.edu) SAN DIEGO SUPERCOMPUTER CENTER Our PSN's first trunk line (to USC/ISI), installed on 29 Oct., was tested this month from here to LA. At last report, however, it was still 150 ft. short of the ISI PSN. The second line to UCLA, service date 4 Dec, is not yet in sight. We are still awaiting 7.4 for our Proteon p4200. We have had no change in our line configuration this past month. by Paul Love (loveep@sds.sdsc.edu) Westine [Page 10]  Internet Monthly Report November 1987 NSFNET REGIONAL AFFILIATED & CONSORTIUM NETWORKS BARRNET BARRNET is operating with reasonable stability of T-1 clear channels, although there is agreement that the telcos need better understanding of sensitivity of non-muxed T-1 channels to the service operations they employ for muxed T-1. Release 7.4 of the Proteon gateway software has been stable at NASA-Ames for over a month now. This version is reportedly 30% faster than release 7.3. Milo Medin has configured the Ames gateways to use "static routes with override", using an initial RIP hop count of 16 (infinity) and then letting RIP dynamically determine which routes are really usable. This allows some administrative control over routing information while still allowing routes to be dynamically marked as up or down. There is consensus that it is desirable to have all BARRNet gateways (routers) run the same configuration. BARRNET's policy and objectives are to achieve maximum connectivity/reachability at the earliest possible time without breaking things. Routing continues to be a problem until the "global" routing metrics are fixed, amongst many other things. Currently most delicate is advertising of routes into BARRNET from other networks through the SDSC gateway and advertising of routes to BARRNET from networks other than those selected. The immediate objective is to advertise to BARRNet from SDSC the networks at the 5 backbone connected supercomputer centers plus the networks at UCSD and UCSB. The SDSC gateway will munge RIP metrics to advertise 1 hop reachability to/from BARRNet and the SCCs. The expanded reachability routing issues (e.g., advertising BARRNet routes outside of the SCCs) will be discussed with NSFNET gurus. For the present, the advertising of these routes must be limited. The BARRNET consortium is now developing its policies for additional membership. About 10 organizations have inquired; government labs, educational institutions and profit sector companies. Early additional connectivity is expected to LLNL, LBL and SLAC. by Bill Yundt (GD.WHY@forsythe.stanford.edu) Westine [Page 11]  Internet Monthly Report November 1987 JVNCNET (Refer to JVNNSC backbone report) MERIT (No report received) MIDNET (No report received) MINNESOTA REGIONAL NET (MRNet) Minnesota Regional Net (MRNet) is waiting on AT&T to install its trunk to NCSA, due December 5th. Proteon P4200 is here. MRNet bylaws approved by charter members. Gated work going on on current ARPANET gateway VAX to allow support for NSFnet, ARPANET and MRNet routing. MRNet technical meetings working on security problems for Cray, ETA, etc. Looks like mail only to them. by Thomas Jacobson (thomas@umn-rei-uc.ARPA) NCSANET NCSA and the University of Illinois Chicago campus will be installing a T1 line to be up by January 1, 1988. The UI-C will serve as a Chicago area hub, allowing area institutions to share T1 access to NCSA and the NSFnet Backbone. A recent meeting was held to discuss issues such as network operations and support as well as to discuss the various possible choices for routers. It was generally agreed that a non-vendor specific hub would be best, allowing institutions to connect to the hub with existing equipment. Seven Chicago area institutions were represented at this meeting. by Charlie Catlett (catlett@ncsa.ncsa.uiuc.edu) NORTHWESTNET (No report received) Westine [Page 12]  Internet Monthly Report November 1987 NYSERNET As of 5 December 1987, NYSERNet had the following topology: Clarkson Syracuse--+ | | | Rochester--------Cornell---------RPI---Albany | || | Buffalo...Fredonia || | | ....Oswego || | | || | Binghamton || +-------- | ------StonyBrook | || | | | || | | CUNY--NYTEL/NSMAC--Columbia======NYU-+ | |\ | | | /| | | | \ NYTEL/GC | NYNEX/S&T / | | | | \ BNL / | | | | \ / | | | | +-------------Rockefeller | | | | | | | +------------------------------+ | | | +-------------POLY-------------------+ || ==== || T1 || \ | / 56kbits .... 9.6kbits The Cornell Columbia/link was upgraded from 56kbits to T1. by Marty Schoffstall (schoff@nisc.nyser.net) SDSCNET (Refer to SDSC backbone report) Westine [Page 13]  Internet Monthly Report November 1987 SESQUINET The complete initially proposed SesquiNet configuration has been operational now for four months. The following campus networks are being served, and are advertised via EGP to the core: Baylor College of Medicine 128.249 Houston Area Research Center 192.31.87 Rice University 128.42 Texas A&M University 128.194 Texas Southern University 192.31.101 and the University of Houston 129.7 During the last month, we have actually experienced one gateway hardware failure and one gateway software failure that effected UH and TAMU, respectively. We have now done some analysis of the reliability of the network, based on logs written by our monitoring program. For each of the sites listed below, we note the number of (unscheduled) failures and the percent of availability (percentage of time the site was up, deducting scheduled down time). -------------------------------- Site Failures Available -------------------------------- BCM 0 100.0 % HARC 5 90.3 % RU 0 99.8 % TAMU 2 99.9 % TSU 2 100.0 % UH 2 99.3 % -------------------------------- While some of the problems were due to leased lines and two gateway failures, most of the unscheduled downtime was due to machine room power problems. The phone line from NSFnet/NCAR to SesquiNet/Rice has been completed, and I am working closely with David Wood of WestNet to put it online using the SesquiNet/cisco router at Rice and the WestNet/cisco router at NCAR. When operational, this link will advertise SesquiNet to NSFnet, and will technically allow us to advertise SesquiNet to USAN and to the NASA Science Network. by Guy Almes (almes@rice.edu) SURANET (No report received) Westine [Page 14]  Internet Monthly Report November 1987 WESTNET 1. A technical workshop was held at the University of Colorado at Boulder from November 9 to 11, 1987. There were attendees from 14 campuses and New Mexico Technet. The first day and one half covered general topics, while the last day was spent on the cisco hardware. The wokshop was very well received by all attendees. 2. Installation of the cisco gateways has been completed at the University of New Mexico and the University of Wyoming. New Mexico Institute of Mining and Technology and New Mexico State University are in the process of being connected. 3. The circuit from New Mexico Technet to NCAR is in the process of being ordered. Installation is scheduld for February 1988. 4. A request for bids is being prepared for all remaining circuits for Westnet. Installation of these additional circuits is scheduled for March 1, 1988. 5. All hardware for Westnet has been ordered, except for the spares, which are planned to be ordered in January 1988. 6. We are preparing a mini-proposal to submit to NSF to connect Brigham Young University and the University of Utah into Westnet. As the circuit to Utah State University already passes through Salt Lake City, the costs involved are minimal. by Pat Burns (pburns%csugreen.bitnet@CUNYVM.CUNY.EDU) TASK FORCE REPORTS ------------------ APPLICATIONS -- USER INTERFACE No report received. AUTONOMOUS NETWORKS The Autonomous Networks Task Force met in Boston, Nov. 5-6 at MIT. Minutes will be available soon. A brief summary of our discussion will be provided in a future Internet monthly report. Deborah Estrin (Estrin@OBERON.USC.EDU) Westine [Page 15]  Internet Monthly Report November 1987 END-TO-END SERVICES The End-to-End Task Force held a meeting at the MIT Endicott House on October 22-23. Host-Group Multicasting The task force chose a target date of July 1988 for moving the IP multicasting spec (RFC966, RFC988) to a proposed Internet standard. First, however, the Task Force must consider some changes which Steve Deering has proposed to the IGMP protocol. These changes would eliminate the Create Group, Join Group, and Leave Group operations, in favor of a mechanism based upon periodic status messages the agent and upon the agents "defending" names in use. The lack of progress towards development of applications using IP multicasting is a disappointment; we hope that moving the protocol towards an Internet standard will encourage its use. Transaction Protocols Jim Stevens of Rockwell described his Transaction Transport Protocol TTP, developed for the SURAN effort. TTP, like the ESP protocol developed at UCL, does not have a strict request/response model, but provides a full-duplex message stream with caching of state information at each end and timeouts to achieve the "connectionless" and minimal-packet properties. The protocol is being implemented at SRI as part of SURAN; it is not known whether this will result in any useful product for research in protocols. One of the major unsolved problems in transaction protocols is how to do flow control without connection setup. It was suggested that the slow-start algorithm of Van Jacobson might be a very good solution. Unfortunately, there are no early prospects for experiments to try out this idea. Performance As always, Van Jacobson stimulated an interesting discussion. He has developed plausible general arguments to show that a stable windowing algorithm must necessarily involve exponential decrease of traffic but linear increase. The argument is based upon the solutions to the first-order difference equation governing a window-based transport protocol over IP. Similar arguments lead to the conclusion that rate-based protocols are potentially unstable. There was some consideration of whether Van's TCP ideas could be applied to dynamic selection of NETBLT parameters. Dave Westine [Page 16]  Internet Monthly Report November 1987 Clark felt that by the time slow-start had found an operating point for the Wideband net, the entire transfer would be over. He plans to work in a different direction, to negotiate with gateways to set up a bandwidth reservation in which NETBLT can work efficiently. Bob Braden (Braden@ISI.EDU) INTERNET ARCHITECTURE Twenty-three requests have been received so far to attend the INARC Workshop on 17-18 December at BBN in Cambridge, MA. The theme of the workshop is the next-generation internet, where we all get to second-guess the Internet architects and build the next one. Since space may be limited, others wishing to attend the bash are invited to send without delay a one- paragraph summary of their favorite agenda topic to: David L. Mills Electrical Engineering Department University of Delaware Newark, DE 19716 (302) 451-6534 or 737-9211 ARPAbox: mills@udel.edu Dave Mills (Mills@HUEY.UDEL.EDU) INTERNET ENGINEERING No report received. INTERNET MANAGEMENT Phill Gross and Vint Cerf have met twice and will meet once more in December to finalize the task force report. Phill is also working on a larger report for Mitre into which much of the result of the task force will go. That report will go to DCA which sponsored the internet planning effort at Mitre. Vint Cerf (Cerf@A.ISI.EDU) Westine [Page 17]  Internet Monthly Report November 1987 PRIVACY The IAB Privacy Task Force had a productive two-day meeting at BBN Communications Corporation in Cambridge, Massachusetts on 4 and 5 November. Attendees were: David Balenson, Curt Barker, Matt Bishop, Morrie Gasser, Steve Kent, John Linn, Dan Nessett, Mike Padlipsky, and Steve Wilbur. A provisional revised version of RFC989 was presented, including several changes and clarifications relative to the released version, and a number of issues were discussed. Authentication/integrity processing will always be applied to the entirety of a message, even when encryption is applied selectively. It was provisionally resolved that the CCITT certificate format will be used to define a public-key certificate's constituent parts. The relation between SMTP addresses and X.400 O/R names is under investigation. We carried out an extended discussion of alternative integrity approaches, in search for a replacement for a single MAC encrypted for each recipient, which Chris Mitchell's papers had identified as inadequate for the multi-recipient case. A per-recipient header parameter will specify an independent choice of message integrity check (MIC) algorithm for each recipient. One option will be the "ambidextrous MAC" LMAC/RMAC algorithm, computing two MACs: one computed in the conventional manner from left to right and one computed from right to left by reversing the order of the 8-octet blocks comprising the message. The key used for these computations will be a variant of the key used for data encryption. We discussed the scope of a future RFC to follow and complement RFC989 by defining the key management infrastructure to support privacy-enhanced mail transmission, and reviewed the state of implementation activities. Subsequent to the meeting, updated documentation was provided to Deborah Estrin of USC to support her organization's proposed work in this area, and meeting minutes were distributed to the membership. Several topics distinct from RFC989 were also discussed. We are checking on licensing issues relevant to use of the RSA algorithm. Correspondence from RSA Data Security, Inc. on this issue was received in response to inquiries after the meeting. We heard a presentation on the British Telecom proposal to CCITT for X.400 electronic mail security. We discussed the Internet Engineering Task Force's concerns about authentication and privacy issues relevant to gateways and monitoring centers. Several messages have been exchanged on this topic, and a follow-up meeting with IETF representatives is anticipated. Westine [Page 18]  Internet Monthly Report November 1987 The next Privacy Task Force meeting is tentatively scheduled for Wednesday and Thursday, 2-3 March 1988, to be hosted by Dan Nessett at Lawrence Livermore National Laboratory. John Linn (Linn@CCY.BBN.COM) ROBUSTNESS AND SURVIVABILITY No report received. SCIENTIFIC COMPUTING Nothing to report this month. Barry Leiner (Leiner@ICARUS.RIACS.EDU) SECURITY No report received. TACTICAL INTERNET No report received. TESTING AND EVALUATION No report received. Westine [Page 19]