[IMR] IMR87-03.TXT Westine [Page 1]  ~ MARCH 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 task forces and contractors in the ARPA Internet Research Program. This report is for research use only, and is not for public distribution. Each task force and contractor 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 ARPANET mail to Westine@ISI.EDU. Reports are requested from BBN, ISI, LL, MIT-LCS, NTA, SRI, UCL, and UDEL. Other groups are invited to report newsworthy events or issues. BBN LABORATORIES AND BBN COMMUNICATIONS CORPORATION --------------------------------------------------- VAX NETWORKING In the month of March, documentation was completed and tapes were made of the final beta release of the 4.3bsd Internet Multicast Westine [Page 1]  Internet Monthly Report March 1987 implementation. Distributions are targeted for Carnegie-Mellon, Stanford and University College London, pending the signing of licensing agreements. (CMU was not able to retrieve the distribution that was made available to them via anonymous ftp last month, due to congestion on the network. They opted to wait for the final distribution, which will be sent via US mail.) WIDEBAND NETWORK BSAT software Release 3.0 was distributed to the Wideband Network sites this month. The Wideband Network's stream service became fully operational with this release, which incorporates automatic detection of and recovery from inconsistencies in the distributed scheduling of satellite channel stream allocations. Also included in Release 3.0 are a a number of robustness features allowing BSATs to operate with only partial host and/or channel connectivity in the event of certain Butterfly hardware failures. Software development efforts were started on two new BSAT features: dynamic group addressing, and shared stream allocations. The former feature allows network hosts to exploit the broadcast nature of the Wideband channel via the dynamic creation, deletion, and membership modification of message delivery groups. The latter feature can provide bandwidth efficiencies for some network applications by allowing multiple hosts to share a single satellite channel stream allocation. BSAT-BSMI hardware and software integration efforts this month resulted in the first satellite channel operation of a BSAT through a BSMI. In addition, BSMI board assembly and testing is well underway, with the first BSMI deployments expected to occur next month. SATNET The SATNET continues to provide reliable service. There have been no additional hardware or software problems. Service was restored on channel 1 at all sites by bypassing the defective piece of hardware in the PSP terminals at Tanum and Fucino. The transmit power has been adjusted at the Fucino site for channel 0 and channel 1. The power drifted and began affecting performance on channel 0. The transmit power is now at an acceptable level on both channels. A long standing and elusive problem on the SATNET has been a slightly higher error rate for particular size packets, "the multiple of 8 bug". The problem was isolated to the transmit encoder in the PSP terminal several years ago with reasonable certainty. We have installed a patch to the SIMP software which pads packets to avoid these dubious sizes. We have tested and Westine [Page 2]  Internet Monthly Report March 1987 installed the patch in the SIMPs but are watching for unknown side effects which it may cause. We no longer see these higher than expected error rates. There has been a connectivity problem in the Fucino SIMP to Cnuce gateway link. The SIMP and Gateway do not appear to be causing this problem. We believe it is a line problem, but we are still investigating. GATEWAYS The Butterfly Gateways continue to be stable. The problem reported last month involving Wideband Gateways' Ethernet Interfaces is under investigation. Work is being done on the Butterfly Gateways Up/Down protocol, to help improve the performance of the ARPANET during congested periods. The European Satnet sites have had problems with connectivity to some Internet sites. Because the gateways do not do reassembly of fragmented packets, EGP routing packets which are too large are being truncated to get them across the SATNET. The European gateways get partial information and may not be able to reach some sites. We are currently working on implementing IP reassembly in the Butterfly Gateway (and LSI-11 too because the problem will also occur when we get to about 300 networks on the ARPANET). Bob Hinden (Hinden@BBN.COM) ISI --- Internet Concepts Project Jon Postel, Bob Braden, and Paul Mockapetris, attended the TCP/IP Interoperability Conference, 16-19 March, 1987 in Monterey, California. Jon Postel gave a presentation on "Internet Protocol Standards", Paul Mockapetris chaired a session for the "Domain Systems" and Bob participated in two sessions and chaired a Birds-of-the-Feather session on network file systems. Jon Postel and Danny Cohen attended the Advanced Computer Communications Workshop sponsored by NSF, at Lake Arrowhead, California, 29-31 March, 1987. Westine [Page 3]  Internet Monthly Report March 1987 Paul Kirton visited ISI on 30 March, 1987, and gave a seminar on his current activities at Telecom in Australia. Paul presented a paper on "Fast Packet Switching for Integrated Network Evolution". Five RFCs were published: RFC 997: Reynolds, J.K., and J. Postel, "Internet Numbers" RFC 998: Clark. D.D., M. Lambert, & L. Zhang, "NETBIT: A Bulk Data Transfer Protocol". RFC 1001: NetBIOS Working Group, "Protocol Standard for a NetBIOS Service on a TCP/UDP Transport: Concepts and Methods RFC 1002: NetBIOS Working Group, "Protocol Standard for a NetBIOS Service on a TCP/UDP Transport: Detailed Specifications RFC 1003: Katz, A., "Issues in Defining an Equations Representation Standard". One ISI research report was published: ISI/RR-87-180: Finn, G., "Routing and Addressing Problems in Large Metropolitan Scale Internetworks". Ann Westine (Westine@ISI.EDU) Multimedia Conferencing Project Our initial tests of packet loss with video traffic have provided mixed results. On several occasions, half an hour or a full hour has gone by without a single packet loss (100000 packets/hour). At other times, there may be a packet loss every 15 seconds to a few minutes. This higher loss rate seems to occur primarily in the BBN to ISI direction, and seems to be more likely after 17:00 PST. We plan to do additional testing over longer periods, and we will try to correlate the video packet loss with other tests of channel bit error rate, etc. Steve Casner (Casner@ISI.EDU) Brian Hung has put aside the addition of text capability in the current multimedia software running on the IBM PC AT, in the interest of making the current program more automated, i.e., one that involves little user effort in sending multimedia messages from the IBM PC AT to the host running the Westine [Page 4]  Internet Monthly Report March 1987 Multimedia Message Transfer Protocol. Joyce Reynolds presented a BBN Diamond demo to Paul Kirton of Telecom, Australia. Brian Hung and Joyce Reynolds (JKReynolds@ISI.EDU) NSFNET Project During March we put in a major burst of effort on the revision to the "Requirements for Internet Gateways" RFC-985. A draft was distributed to about 15 people, and we received and processed very thorough and useful comments from Hans-Werner Braun, Scott Brim, Dave Mills, Craig Partridge, and Steve Wolff. Our goal is to release an advanced draft to the Internet Engineering Task Force for comments before their April meeting at BBN. There are still a number of important voids in the gateway specifications; we look to the Engineering Task Force to take an active role in filling these voids. We also continued work on gathering data on the current NSFNET topology. Our goal is to design a database on network connectivity for use in future planning and operations, and to create an initial instance of that database. If we collect the right data, we will be able to extract from it "views" with appropriate levels of detail for particular purposes. For example, we want to be able to suppress or include the inner structure of the networks, and to suppress or include the subnet details. As a byproduct of this work, we prepared a one-page map of the current NSFNET, at the lowest level of detail; copies of this map found their way into a number of talks at the Monterey TCP/IP Workshop. Bob Braden and Annette Deschon attended a half-day meeting on NSFNET technical issues, held at Monterey in connection with the TCP/IP Workshop, March 16-19, 1987. Braden attended the workshop to participate in two sessions and to chair a Birds- of-the-Feather session on network file systems. Deschon attended a two-day meeting of the re-constituted Autonomous Networks Task Force, chaired by Deborah Estrin, 20-21 March 1987, at Xerox Parc, in Palo Alto, CA. Bob Braden (Braden@ISI.EDU) Supercomputer and Workstation Communication Project We have finally achieved NETBLT transfer rates near the throughput limit of the Wideband Net. The overall rate (including initial and final handshaking) for a transfer of 716000 bytes was 521 Kb/s. But for the purposes of testing Westine [Page 5]  Internet Monthly Report March 1987 the network, the more interesting measure is the steady-state transfer rate which ignores the initial and final handshaking and any retransmissions after the initial transmission of the last packet. The steady-state transfer rate is equivalent to the overall transfer rate as the file length approaches infinity. The best steady-state transfer rate was 942 Kb/s, nearly hitting the magic goal of 1 Mb/s! The theoretical transmission rate for the data alone was 1.04 Mb/s, but packet loss was the cause of the lower measured value. The loss is not in the Wideband Net, but appears to be in reception at the destination Sun. Steve Casner (Casner@ISI.EDU) MIT-LCS ------- NETBLT testing is continuing over the 3 Mbps Wideband satellite network. We have reached throughputs of 945 Kbps, out of a maximum usable bandwidth of a little over 1 Mbps. Attempting faster transmission rates has caused the Wideband net to fail in various ways, due to overload of either the BSAT or ESI network interfaces. The next phase of NETBLT testing will involve the design of a series of adaptive rate-control algorithms to provide automatic selection of optimum transmission rates as the load over a particular network path varies during a data transfer. Lixia Zhang (Lixia@XX.LCS.MIT.EDU) NTA and NDRE ------------ 1. The Butterfly gateway and SATNET have been working stable for months. Morning connectivity to the West Coast (our time) is not good but acceptable. Afternoon connectivity is useless! 2. The area where NTA is located has changed to a new modern digital telephone exchange, with the result that all numbers in the area have been changed. For those in need of it, the new number for NTA-RE is (+47) (6) 80 91 00 or (+47) (6) 80 91 25 3. There is no third spare modem in the PSP terminal at Tanum, which worries me and Knut Lindquist quite a bit. Allthough everything seem to work well right now, the setup at Tanum has Westine [Page 6]  Internet Monthly Report March 1987 shown considerable flakyness from time to time. There is very little info like schematics etc on the PSP terminal, thus very little help for Knut when things are going bad. Only his enthusiasm, strong feeling of responsibility and hard work has made Tanum work so well as it does. What can be done to improve the situation? 4. We have changed to the new domain-style adresses. NTA-VAX.ARPA is renamed to TOR.NTA.NO NTA-ODIN.ARPA will shortly be renamed to ODIN.NTA.NO 5. NTA-RE has started a multi-media activity which will be based on Diamond. People have been allocated and are awaiting the final go from Darpa and BBN. 6. There seems to be substantial interest at NTA-RE to experiment with new satellite services. Probably this could be combined with the ICB interest. A roof-top setup (antenna and RF equipment) made in Norway would be approximately Nok 250.000.-, and about the same as I pay for the floor-space at Tanum and 48 kb/s line from Kjeller to Tanum! The best of all, we can do away with the Swedish customs people! 7. NDRE is currently actively using ARPAmail as a means to communicate with various NATO groups and others. Hopefully this can stimulate to further involvements. Paal Spilling (paal@TOR.NTA.NO) SRI --- No report received. UCL --- Steve Kille attended the second TCP/IP Interoperability Conference at Monterey, CA, and gave a talk on X.400 and using the ISODE. Steve Easterbrook finished the preliminary implementation of the C ASN code generator generator(!). Work has started on a tool to automatically draw reasonable pictures of networks, given lists of gateways and attached networks (ie host.txt file). This will be integrated in the UCL status and alarm system. Jon Crowcroft (jon@CS.UCL.AC.UK) Westine [Page 7]  Internet Monthly Report March 1987 UNIVERSITY OF DELAWARE ---------------------- 1. Development continues on the Dissimilar Gateway Protocol (DGP). A technical note on data-base design and transmission model was distributed. A simple link-based routing algorithm was designed and a test jig built (in BASIC, no less). The test jig used real data collected from NSFNET and its clients, consisting of 78 gateways, 63 nets and 217 interfaces interconnected in glorious ways. The Linkabit folk are working on a specification due for review in April. 2. The University of Delaware hosted a one-day, ad-hoc meeting on high-speed networks with several luminaries from NRI, Bellcore, MIT, AT&T Bell Labs, IBM and UDel in attendance. The group, so-far mercifully unnamed, agreed to meet at Bellcore next month. 3. Gary Delp, Dave Farber and Dave Mills attended the Advanced Communication Workshop at Lake Arrowhead, CA. Pete Warter and Dave Mills attended the TCP/IP Interoperability Conference at Monterey, CA. Dave Mills attended a NAS meeting on survivable telephone networks at Washington, DC. 4. A new fuzzware version for NSFNET Backbone sites and other places has been completed and is now in test. Besides fixing a few bugs, the new version has much more buffer space due changes in the virtual-storage mapping mechanism. Linkabit moved their fuzzballs and Suns to another location in the building, with the result that the WWVB radio clock is currently antennaless and its indications senseless. In spite of that, clockwatchers throughout the Internet still hear high-quality time signals, since the intricate system of NTP- based backup clocks keeps on tocking. Dave Mills (Mills@UDEL.EDU) Westine [Page 8]  Internet Monthly Report March 1987 NSF NETWORKING -------------- UCAR/BBN LABS NSF NETWORK SERVICE CENTER (NNSC) We continue to add to the services and information we provide the NSFNET community. We are coordinating the NSFNET contribution to the Internet Monthly Report, and have attended some recent meetings: Dick Edmiston and Karen Roubicek went to the Federation Assembly and FCCSET Meetings in San Diego, and Craig Partridge gave several presentations at the TCP-IP Interoperability Conference in Monterey. We are working on developing network monitoring standards and Craig expects to present a proposed system to some of the Internet Task Forces in April. Craig, Karen and Bill Curtis from UCAR visited the Pittsburgh Supercomputer Center and met with the Systems, User Services and Operations Managers to get an overview of the Center's operations. We have now visited five of the six centers. We have published a bulletin to increase the user community's awareness of NSFNET and to give them information which will make it easier for them to use the network. The bulletin will be mailed out the week of April 6. By Karen Roubicek (roubicek@nnsc.nsf.net) NSFNET BACKBONE SITES CORNELL NATIONAL SUPERCOMPUTER FACILITY 16,800,000 packets were delivered (net) by the backbone in February, up 35% from January. There was a surge at the end of February. At the end of the first three weeks of March we had delivered 12,000,000 -- 65% more packets than the same period in February. We are now monitoring and collecting traffic data for SURANET's connection to the NSFNET backbone. If you would like access to this sort of information contact the Network Information and Support Center at 607-255-5060. Coordination and Interoperability Issues Cornell Theory Center staff are now participating in the NPAG, four of the IAB task forces and two of the the NSFNET federation committees. There have been no major changes in the gatedaemon code, although some bugs were fixed. We are now centrally coordinating all of the gatedaemon implementations that surround the backbone, including Westine [Page 9]  Internet Monthly Report March 1987 the EGP metrics with which networks are announced to ARPANET. By Scott Brim (SWB@DEVVAX.TN.CORNELL.EDU) UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN We had planned to drop off of the USAN network in preparation for turning into our own hub at the time NCAR converted their addresses. This would not have been a problem with Illinois users, but would have been a problem for Northwestern and U of Chicago. Therefore, for the time being we are still part of USAN. TCP et. al. for CTSS is nearing completion. Next week we will begin staff testing a FTP to/from CTSS. Telnet will come a bit later. Performance is comparable with our current Move utilities even using the bottleneck Interprocess Communication facility in CTSS. Slicker IPC facility is currently being tested. Class B conversion of campus is immenent. Our P4200's have all been converted to 7.2 clearing the critical path for the conversion. We will need to change our fuzz address probably to 128.174.5.14. Date for conversion should be set Monday with an estimated two weeks of user warning. By Ed Krol (krol%uiucvmd.bitnet@WISCVM.WISC.EDU) JOHN VON NEUMANN SUPERCOMPUTING CENTER Dynamic Routing: Routing at the John von Neumann Center ( JVNC ) is done dynamically with RIP between "some" of the members of the JVNC consortium and HELLO with the other NSFNET sites. We are currently working on "integrating" the campuses networks into the JVNCNET. This means: 1. managing the routing information so any campus in the consortium can get to any other campus of the consortium via JVNCNET and/or any other alternative path, 2. pass to the consortium sites the routing information learned from NSFNET core system, 3. filter any bogus network from being advertized to the NSFNET backbone, 4. provide full backup routing to the members of the JVNC Consortium so they can route to the ARPANET if their PS is down. Some of the Consortium sites's networks are already fully integrated to the JVNCNET. These are: Westine [Page 10]  Internet Monthly Report March 1987 Rutgers University, University of Pennsylvania Princeton University, Columbia University, New York University, Penn State University, University of Colorado. The other sites are still working out local network problems to make this integration possible. Note that JVNCNET will soon provide alternative paths to the NSFNET backbone since, 1. University of Arizona is going to be connected to the San Diego SuperComputer Center, and they are already connected to JVNC via JVNCNET. 2. University of Colorado is already connected (although not exchanging dynamic routing information) with NCAR, and they are connected to JVNC via JVNCNET. The University of Colorado is logically connected to the JVNC-NSFNET core gateway via a VITALINK bridge over the satellite. 3. New York University, Columbia University and University of Rochester are all part of NYSERNET (as well as the Cornell Theory Center), and they are also part of the JVNCNET. University of Rochester is expected to start exchanging dynamic information with JVNC in a few days. NYU and Columbia are already doing so. NYU is connected to JVNC via a T1 line while Columbia and Rochester are connected with 56k lines. These multiple connectivity paths will add accessibility from the different Universities to the Supercomputer Centers, and redundancy to the NSFNET backbone. Network Reliability: The network is very stable these days, occasionally the TransLAN bridge that connects our satellite nodes hangs, but VITALINK is working on the problem. The T1 lines are very reliable and if they go down we have them tested and back up in record time. The 56k lines are reliable as well, with the exception of the link to Columbia, the problem at the New York site was to convince NYtel that the CSU/DSU was good, finally they found the problem in their point of presence in Columbia. We are testing Ungermann-Bass routers (we have 6 of them). They are on alpha test right now, and connect JVNC with NJIT, NJIT with Stevens and NJIT with UMDNJ. They have been running for a few weeks already. Westine [Page 11]  Internet Monthly Report March 1987 We are increasing the traffic on our T1 lines as the scientists at the different sites start to find out about X-Windows. We expect to have full utilization of the T1 bandwidth as graphics users grow. Network Monitoring: Together with the emergence of backbones, regionals and other entities, a need came about to monitor the networks. At JVNC we are working on a project that is now in phase II, to monitor the JVNCNET network and the NSFNET network as well. This project consists on an integrated package that consists of: a. a scheduling program to schedule up/down time, b. a monitoring program to collect interfaces statistics, as well as gateways dynamic routing information, c. reachability program, to determine the networks that are reachable, and help us find routing loops The programs have user interfaces in the form of color graphic displays in our user services area, status reports for the technical staff and for the managers, and reports for the operators to indicate problems on the network. By Sergio Heker ("heker@jvnca.csc.org") NATIONAL CENTER FOR ATMOSPHERIC RESEARCH AND UNIVERSITY SATELLITE NETWORK PROJECT A problem with a switch on the LSI11/73 caused the fuzzball to be down for about 24 hours on March 30. The splitting of address space 128.117 into 128.116 and 128.117 at NCAR has been completed. The fuzzball to NSFNET (X.64.3), cisco gateway to the University of Colorado (X.64.4), Proteon gateway to 128.117 (X.64.2), and all USAN gateways { UMD (X.6.1), Wisconsin (X.12.1), Illinois (X.4.1), Oregon State (X.14.1), Michigan (X.18.1), Miami (X.10.1) } are now on X = 128.116 while all NCAR hosts are isolated on 128.117. The Proteon gateway to 128.117 is configured with static routing tables with the default gateway being the fuzzball. On April 7 Illinois will drop out of USAN but will still be accessible to NCAR via NSFNET. Woods Hole will be added to USAN sometime in May. Since Woods Hole also is connected to MIT, it may have a backdoor path to NSFNET via JVNC. The NCAR system now looks like Westine [Page 12]  Internet Monthly Report March 1987 Satellite to JVNC | Univ. of Colo. satellite to USAN | sites with Vitalink fuzzball to cisco box TRANSLANS NSFNET | | | 128.116 --------------------------------------------------------------- | | PROTEON p4200 {this summer} -> Sun 3 -- ARPA PSN | | --------------------------------------------------------------- All machines at NCAR 128.117 By Don Morris (morris@scdsw1.ucar.edu) PITTSBURGH SUPERCOMPUTER CENTER NSFNET The PSC Fuzzball has had few interruptions. It was rebooted on March 25th so that it is now running the latest release of fuzzball software. The UIUC-PSC link has been upgraded by Illinois Bell to a direct 56Kbps DDS circuit. We had previously announced to NSFNET-OPS our plan to relocate the lines for our Fuzzball to the building where our offices are. We now have a commitment from Bell of Pa to do the line relocation on Monday, April 6. This involves moving PSC Fuzzball and PSC-GW to another building, reconnecting the lines to JVNC and UIUC, and extending the line from PSC-GW to the CMU PSN. PSC-GW has been running quite well. It is currently averaging about 465,000 incoming packets per day. Beta tests of a new release of ACC 5250 software and of flow control modifications to our local PSN have been completed. These tests have greatly improved the throughput and reliability of the NSFNET/ARPANET link. PSCNET We have had problems with IP traffic to our Front End 8650s (VMS v4.5) from both NSFNET and ARPANET. We have felt that most of the speed problems and some of the disconnect problems were due to our local routing: NSF traffic comes into our Fuzzball, to PSC-GW over a stub Ethernet, out PSC-GW to the CMU PSN, through the PSN and CMU-GW to a heavily loaded 3mb Ethernet, through an 11 router with multiple drops to a 10mb Ethernet, through a locally designed 68000 router to the CMU Campus spine, through a DEC LANBridge to our offices, and from there through a Vitalink Translan II using a T1 line to our remote facility where our Cray and Front Ends are. At the end of February, we installed a second network interface in CPWSCB to provide an alternate channel for experiments in improved IP service to our Front Ends. This Ethernet controller was connected to a stub Ethernet whose only other occupants were a Proteon P4200 and its boot device. Westine [Page 13]  Internet Monthly Report March 1987 This 4200, located at our remote machine room was connected to another 4200, located at our offices, which was in turn connected to a 4200 at the site of our gateway, fuzzball, and PSN. (The link from front ends to offices is a portion of a T1 channel, about 750kbps; the link from offices to Fuzzball is 56kbps. The 56Kbps link will disappear when the fuzzball is reterminated as described above.) This has enabled us to bypass the multiple routers and use an unloaded link. We went through several configurations of interface on the 4200s ("loads"), until, with Proteon's assistance, we have achieved uptimes on the order of 5-10 days without crashes. Even during periods of instability, users have been pleased with the improved performance. We are now announcing service to B.PSC.EDU exclusively at 128.182.65.2, whereas it was previously in the CMU net (128.2). We plan to move A.PSC.EDU to this new net, leaving it dual-homed, if possible. This has given us a good shakedown of the P4200s, so we're planning to start deploying them to build up PSCNET straight away. Several lines have been ordered and the links will be brought on line as soon as the vendor can provide them. By Jim Ellis (ellis@MORGUL.PSC.EDU) SDSCNET SDSC has put in place a Proteon p4200 to server as an IP gateway between its Ethernet and the UCSD campus-wide broadband/ethernet. A DEC LAN Bridge 100 is in parallel to the p4200 to provide DECNET and DEC LATS connectivity between the 2 Ethernets. This has been functioning well since mid Feb. The Proteon will also connect our first "TCP/IP" consortium member, UC Santa Barbara, to the center at 56k. This is expected to be operational as soon as line problems are resolved. We have ordered a T1 interface for the p4200 to use with a microwave link to the Salk Inst. This will provide them with IP, DECNET and BITNET service. The p4200 will also be used to link an industrial participant to the center at 56k with both IP and DECNET. SDSC has ordered a center-funded 56k circuit to UC Los Angeles to provide a connection directly to the ARPANET. This service, provided via the UC Office of the President (UCOP) T1 service between UC San Diego and UCLA, will be used until the SDSC/ARPANET PS, scheduled to be installed in Aug, is connected by DCA lines. As a side note, UCOP service is also used to connect the center to UCSB. Westine [Page 14]  Internet Monthly Report March 1987 The center has taken delivery of a DEC 8200. This will be used as a research tool to explore methods of high speed data flow between our hub and the Cray X-MP/48; speeds higher than that currently provided by our NSC Hyperchannel. NSFNET REGIONAL AFFILIATED AND CONSORTIUM NETWORKS BARRNET (No Report Received) JVNCNET (Refer to JVNNSC backbone report) MERIT The Merit Computer Network is currently focusing most of its development efforts on implementing the TCP/IP protocol suite within the network and its host interfaces. In March, progress toward that goal included: + Software internal to the subnet for control of IP-packet congestion. Enhancement of routing functionality. Proliferation of routing tables throughout the subnet. ARP source IP address verification. + At the subnet boundary, the provision of automatic recognition of the asynchronous IP protocol on an integrated basis with our existing terminal-recognition procedure. Provision of a UDP service access point for Merit channel-attached hosts. + At the user level, the first user beta-testing of our IP services. Beginning of campus-wide training of user-services staffs in what will be required for support of the TCP/IP world. By Christine Wendt, Merit Technical Support Group (Christine_Wendt@UM.CC.UMICH.EDU) MIDNET The big news for the MIDNET institutions (the Univ. of Nebraska- Lincoln, Kansas State Univ, the Univ of Kansas, the Univ of Tulsa, Oklahoma State Univ, the Univ of Oklahoma, the Univ of Arkansas- Fayettevill, the Univ of Missouri-Columbia, Washington Univ, the Univ of Iowa and Iowa State Univ) is that we have been able to get 56 kbps digital service into all locations. We have been promised a June 1 installation of the following configuration: Westine [Page 15]  Internet Monthly Report March 1987 UNL --- --- ISU --- UI --- NCSA | | KSU --- KU | | UMC --- WU --- | | OSU --- UT | | | OU --- --- --- UA In addition to the current members, the consortium is exploring connections to additional universities that have expressed an interest. On March 30 representatives from the MIDNET institutions attended a one day seminar at the National Center for Supercomputer Applications. The seminar covered the facilities and services offered by the NCSA as well as an introduction for first-time supercomputer users. We hope that this will be the first of many meetings aimed at the end user. The consortium has initiated a publications/documentation program that will organize its memoranda and technical notes and start a monthly newsletter. The first issue of the newsletter is scheduled for May. Doug Gale (doug%unlcdc3.bitnet@WISCVM.WISC.EDU) NORTHWESTNET Northwestnet has not yet been formally funded; however, we are optimistic that final NSF approval will be forthcoming within the next 30 - 60 days. Current plans are to connect the following universities via 9.6/56 kbaud lines: University of Washington, Seattle, WA Washington State University, Pullman, WA University of Idaho, Moscow, ID University of Alaska, Fairbanks, AK Montana State University, Bozeman, MT North Dakota State System, Fargo, ND University of Oregon, Eugene, OR Oregon State University, Corvallis, OR Oregon Graduate Center, Beaverton, OR Connection to the NSFNET has not been decided upon, but the University of Washington has a 56kbaud satellite link to the San Diego Supercomputer Center, and Oregon State University has a VITALINK Satellite connection to NCAR. Westine [Page 16]  Internet Monthly Report March 1987 The three principal investigators are Robert McVicar, Northwest Academic Planning Forum, Dick Jonsen, Western Interstate Commission on Higher Education (WICHE), and myself. The network will be managed by a management committee, having one representative from each university, plus one from Boeing Computer Services (BCS). BCS will be the facilities manager for the network. Technical supervision will be provided by a Technical Committee, chaired by Dr. John Sobolewski, University of Washington. By Hellmut Golde (Golde%UWACDC.Bitnet@WISCVM.WISC.EDU) NYSERNET As of 1 April 1987, NYSERNET has the following topology with 56kbit links. T1's are also in place for the topology of below but have not been cutover to the switching gear. Packet traffic from NYSERNET, NSFNET, ARPANET etc is available to each site. Rochester--------Cornell---------RPI | | | | | | NYTEL------------Columbia--------NYU | | | BNL (Brookhaven National Laboratories) "Line" facilities are in place for SUNY Albany, SUNY Buffalo, SUNY Binghamton, SUNY Stony Brook, Polytechnic University, City University of NY, and Rockefeller University. Delivery of preconfigured switching gear for these will commence in the 2nd week of April. Planning for two corporate members, IBM and Kodak is in progress. Creation of an experimental root domain server is in progress. A meeting restricted to TCP/IP Host Implementation and interoperability is being planned for all "systems people" within the NYSERNET community for the 27th of April at the University of Rochester. By Marty Schoffstall (schoff@NIC.NYSER.NET) SDSCNET (Refer to SDSC Backbone Report) SESQUINET Our immediate goal is to have an initial working configuration up by June 15th. This initial configuration will include all members involved in the NSF proposal that funds the network: Westine [Page 17]  Internet Monthly Report March 1987 Baylor College of Medicine Houston Area Research Center Rice University Texas A&M University Texas Southern University University of Houston, Central Campus The initial topology will be only slightly redundant: TAMU \ HARC \ / \ / UHD / \ (ARPANET)-----RICE UHCC | | BCM-----TSU Each of the links is a 56-kb/s DDS circuit, with the exception of the HARC-to-UHD link, which will bundle several 56-kb/s circuits into a single higher speed link. All the site abbreviations should be clear with the exception of the switching router (supporting no local campus network) located physically at the Downtown campus of the University of Houston for convenience of telecommunications. Initially there will be one 56-kb/s link to the rest of the internet, viz. the existing ARPANET connection to Rice. We are this week ordering the telecommunications links, and expect them within two months. Early this month we will decide on router vendor; we are most seriously considering Cisco and Proteon. Later this summer we expect to connect to the NSFNET site at NCAR. Further, we hope to connect to a SURANET site. The inclusion of these links will bring to three the number of medium-speed links to the rest of the internet. Although we are concentrating on implementing the initial configuration, we are always interested in adding members. Potential members should contact Guy Almes at (713)527-4834 or almes@rice.edu. Guy Almes (almes@RICE.EDU) SURANET Eleven SURANET routers are presently interconnected by 56kb lines and are operational. A twelfth should be operational within a few Westine [Page 18]  Internet Monthly Report March 1987 days. Four of the sites have their campus network on-line to the router. All sites use Proteon p4200 IP routers. At the time of the last report all used Version 7.1 of the Proteon software. Reliability was poor. During the past month we have moved to Version 7.2 and have installed a simple hardware modification (100 ohm pull-up resistor ). Reliability appears to be much improved. During this month we began to advertise the following nets via EGP to the internet: 128.109.0.0 TUCC 128.150.0.0 NSF 128.163.0.0 U of Kentucky 128.164.0.0 George Washington Univ 128.167.0.0 SURANET 128.169.0.0 U of Tenn 128.173.0.0 Va Tech 192.5.57.0 Univ of Delaware (udel-cc) 192.5.219.0 Clemson 192.16.17.0 Univ of Alabama We are working on improving the reliability of the fuzzball which communicates with Cornell. Operator procedures and training are being developed so that 24 hour coverage will be available. Operator procedures for the Proteon routers are also being developed. Drs. Hahn and Ricart, as well as four other University of Maryland staff participated in the Interoperability Conference, Mar. 16-19, Monterey, CA. A SURANET technical meeting was held in mid-February at the University of Maryland. Over 70 people attended from SURANET and from other NSF supported networks. The meeting was videotaped and copies can be purchased. The first of a series of user's meetings is planned for next month. The host will be Dr. Henry Schaffer of the University of North Carolina. Attendees from non-SURA institutions will be welcome and should contact Dr. Schaffer at (919) 737 2516. By Jack Hahn (hahn%umdc.bitnet@WISCVM.WISC.EDU) WESTNET No report received. Westine [Page 19]  Internet Monthly Report March 1987 TASK FORCE REPORTS ------------------ APPLICATIONS -- USER INTERFACE No report received. AUTONOMOUS NETWORKS Several members of the Autonomous Networks Task Force met March 20 and 21 at Xerox PARC. I am now working on writing up the meeting notes as a basis for ongoing discussion over the net. Persons interested in participating in the discussion should ask to be added to the autonets-interest@isi.edu mailing list (send your request to me, estrin@usc- cse.usc.edu). Our next official meeting is scheduled for early November in Boston. Deborah Estrin (Estrin@USC-CSE.USC.EDU) END-TO-END SERVICES No report received. INTERAGENCY REQUIREMENTS The first meeting for Interagency Requirements will be in May. Vint Cerf (Cerf@A.ISI.EDU) INTERNET ARCHITECTURE The INARC list was quiet this month. Dave Mills (Mills@UCEL.EDU) INTERNET ENGINEERING No report received. Westine [Page 20]  Internet Monthly Report March 1987 PRIVACY Privacy Task Force minutes from the September 1986 meeting were distributed this month, in order to facilitate continuing discussion of electronic mail key management mechanisms at our next meeting, which will be held 31 March-1 April at RIACS. John Linn (Linn@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. Westine [Page 21]