[IMR] IMR87-07.TXT Westine [Page 1] ~ JULY 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 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@SH.CS.NET). BBN LABORATORIES AND BBN COMMUNICATIONS CORPORATION --------------------------------------------------- Wideband Network The Wideband Network monitoring protocol and the BSAT software have been modified to accommodate simultaneous network monitoring by multiple host computers. Without these modifications, Wideband monitoring could only be performed by a single host located at the Network Operations Center. The multisite monitoring support will be included in the next BSAT software release. Work started during the month on the integration of a user-code host interface device reset function into the BSAT. This function Westine [Page 1] Internet Monthly Report July 1987 obviates the use of full system restarts to correct certain problems in the host interface devices. The full system restarts have the undesirable side effect of deleting local stream allocation information. A coast-to-coast meeting of the End-to-End Task Force was supported by the Wideband Network and the multimedia conferencing facilities at BBN and ISI on July 29. SATNET The SATNET continues to remain stable. Measurements by ISI showed an availability of 99% for all the SIMPs except Tanum. The Tanum SIMP had a memory board burn out. A field service engineer was dispatched to the site and a new board and fan were installed. The availability of the Tanum SIMP was 96%. INTERNET R&D The number of networks in the Internet has passed the 250 mark. As a result the EGP reachability messages sent over the ARPANET are (almost) full. We plan in the near future to begin updating the software in the LSI-11 and Butterfly Gateways to fragment the EGP messages. Gateways which run EGP with the core gateways will need to support IP Reassembly. If you have a gateway which this will be a problem, please contact us (gateway@bbn.com) as soon as possible. We are implementing support in the Butterfly Gateway for the parts ICMP of not previously supported, notably Source Quench. We are including the capability to answer source routed ICMP echos via the reversed route. We also found and fixed a bug relating to the handling of strict source route. Our implementation of the SURAN LPR interface is progressing. We expect to begun debugging with a real LPR in August. We are tracking a bug which is causing the European Satnet Gateways to restart. We apparently, introduced this bug when we added the IP reassembly code in June. We hope to find and fix it soon. We are sponsoring a meeting at BBN to get comments on the Routing proposals currently before the ANSI committee discussing IS-IS routing. We hope to present comments from the Internet community. If interested in this, please contact Ross Callon (rcallon@bbn.com). Bob Hinden (Hinden@BBN.COM) Westine [Page 2] Internet Monthly Report July 1987 ISI --- Internet Concepts Project Jon Postel and Bob Braden attended the IAB meeting in San Francisco, July 7-9. Two RFCs were published. RFC 1005: Leiner, B., "Implementation Plan for Interagency Research Internet". RFC 1006: Prue, W., and J. Postel, "Something a Host Could Do with Source Quench: The Source Quench Introduced Delay (SQuID) Greg Finn continues to work on the issue of attack resistant routing procedures. Ann Westine (Westine@ISI.EDU) Multimedia Conferencing Project Work continues on the implementation of packet video protocols for conferences with more than two sites. Meanwhile, there was a flurry of teleconference activity at the end of July using the existing two sites at ISI and BBN. The first teleconference meeting of the End-to-End Task Force was an all-day affair similar to previous tele-meetings of the Internet Activities Board. The End-to-End Task Force meeting relied on MMCONF/Diamond for presentations of "viewgraphs" much more than the IAB meetings have. There was just one short interruption of the packet voice and video; we had more trouble with the MMCONF connection, largely due to repeated gateway outages that left the host with a bad route since there's no automatic mechanism yet to try another gateway. The participants returned several valuable comments on how the teleconferencing system could be improved. A shorter teleconference was held for members of a pair of cooperating projects at ISI and BBN working on a circuit design. MMCONF allowed shared viewing of a diagram with each site pointing out various features. The participants seemed to like the conferencing system since they asked to use it again in two weeks. We wrapped up the month with demos for Gordon Bell of NSF and Ed Brown of DARPA. Westine [Page 3] Internet Monthly Report July 1987 Brian Hung has written a shell script program to output a bitmap file in RFC-797 format from a Sun workstation to an Imagen printer. The program is made up of three separate programs, expand, tosun and sunimp. The program expand was written to expand bitmap files by one and a half times in the X and Y coordinates. This is required to make the printed hardcopy the same size as the original due to the fact that the Microtek MS-200scanner has a 200 dots per inch resolution while the Imagen printer has a 300 dots per inch resolution. Steve Casner and Brian Hung (Casner@ISI.EDU, Hung@ISI.EDU) NSFNET Project The background file transfer daemon program was released for testing within ISI, and preparation of an RFC on it was begun. Planning began on a statistics-collection facility that will allow the NSFNET backbone and regional networks to gather a variety of kinds of statistics on the level and nature of the traffic. Work continued on the SUN version of NETBLT. Using Mark Lambert's revised application interface, we were able to operate NETBLT as the data transfer protocol under the BSD versions of user and server FTP. Bob Braden attended a two-day meeting of the IAB, and ran a one-day video teleconferencing meeting of the END2END task force at ISI on July 22 (see report below). Considerable effort was consumed in organizing the NSFNET Technical Workshop to be held at Cornell University at the end of August. Bob Braden and Annette DeSchon (Braden@ISI.EDU, DeSchon@ISI.EDU) Supercomputer and Workstation Communication Project Alan Katz continued work on the Intelligent Communication Facility reported on last month. Alan also continued experimentation with X-Windows and with the newly released X toolkit from DEC. This toolkit is a V10 implementation of their toolkit for V11. Alan packaged up the files which create his suntools-like X environment into a sharfile for others to try. To access it, FTP the file: /div2/katz/myx.sharfile on VENERA.ISI.EDU, Westine [Page 4] Internet Monthly Report July 1987 unshar it using sh, and follow the directions in the README file. This will only work on Unix systems and has only been tested on a Sun-3, but it should work with other workstations. It is very experimental, but should provide a way to start up X without wading through a lot of documentation. Alan attended SigGraph'87 July 29-30, in Anahein, California. Alan Katz (Katz@ISI.EDU) MIT-LCS ------- No report received. MITRE Corporation The objective of the MITRE Internet Engineering program is twofold: 1) to address internet level performance issues for the DoD internet and 2) to address the interoperability between the DoD and OSI protocol suites as support for a planned transition from DoD to OSI protocols. To support these objectives, the following work was accomplished this month: 1. Internet Performance Measurements The performance tests exercising multiple network interfaces received a more accurate interval timer, adjustable window size in the transport layer, and expansion of hops to include multiple gateways. 2. Congestion Control Experimentation We are comparing results from simulated congestion control experiements with the internet performance measurements extracted from the ARPANET and MILNET. 3. FTP/FTAM Application Bridge Bidirectionsal traffic is being tested using FTP compatible functions as the test suite. 4. VTP We completed the negotiated release function in the prototype and continued work on the design and implementation plan. At the NBS Implementors Workshop, the VTP SIG, the TELNET profile was approved for VTP Phase 1a, which contains TELNET Westine [Page 5] Internet Monthly Report July 1987 compatibility. 5. Name Service In coordination with the NIC, at the Internet Engineering Task Force, we reviewed the RFC's for Name Domain in the MILNET. 6. Gateway Monitoring We began implementation of an instrumented DoD internet gateway and an associated remote monitor. We also extended the basic design to address an instrumented dual protocol (DoD and OSI) internet gateway. 7. Standards a. NBS Implementors Workshop, FTAM SIG There were two major topics discussed: 1) the expanded use of character sets in diagnostic messages to support European characters not covered by the IA5 character set; and 2) the use of Version 1 session specifications for Phase 2 FTAM implementations. Version 2 specifications will apply to Phase 3 FTAM. b. NBS Implementors Workshop, VTP SIG The TELNET profile for use in VTP Phase1a was approved. c. ANSC X3S3.3 DEC and UNISYS presented proposals for IS-IS routing within a single domain. Ann Whitaker (whitaker@GATEWAY.MITRE.ORG) NTA & NDRE ---------- No report received. Westine [Page 6] Internet Monthly Report July 1987 SRI --- Internet Research Jose J. Garcia-Luna-Aceves attended the Applications Task Force meeting at Stanford; Jim Mathis attended the Internet Activities Board meeting in San Francisco; and Zaw-Sing Su attended the ANSI X3S3.3 (Transport and Network Layer Working Group) meeting in Bellevue, Washington and Internet Engineering Task Force meeting at MITRE Corp., Mclean, Virginia. We are sorry to report the departure of Jim Mathis from SRI to join the commercial sector. His contributions to the Internet community will be missed. Zaw-Sing Su (ZSu@KL.SRI.COM) Multimedia Conferencing/Wideband Network Under the Command and Control Workstation Project sponsored by the Naval Ocean Systems Center, we are investigating the use of Wideband Satnet as a communication link to support multimedia conferencing over the long haul. As a first step in evaluating the network, we created a testbed to examine the effect of network delay on conferencing, the practical problems of connecting to the network, etc. This testbed consists of a "loopback" route between SRI and ISI, two Suns at SRI acted as conference workstations, and a third Sun located at ISI rerouting any packet it received from one SRI Sun, to the other. We gave both Suns at SRI the ISI Sun's address as the other participant in the conference, and forced them to use the Wideband network by fixing their routing tables. Traffic from one SRI Sun went to our Wideband gateway, over the satellite network to ISI, and through their gateway to their Sun. Their Sun sent the data back through their gateway, over the satellite network again, through our gateway a second time, and then to the other SRI Sun. Clearly, we were testing a "less-than-realistic" situation. The packets not only traveled over the Wideband network twice, but also went through another gateway and Sun. However, it allowed us to perform the test with both ends of the conference at the same site, which made testing easier. The testing was successful. The network delay did not appear to interfere with conferencing activities in general, although it made explicit transfer of the floor necessary (similar to using a two-way radio). After a few false starts, we found it easy to configure a workstation to use the Wideband network. Our next experiment will be a conference over the Westine [Page 7] Internet Monthly Report July 1987 Wideband network between two separate sites. Keith Williams (Keith@tsca.sri.com) Distributed Applications Under the ADCCP (Advanced Distributed Command Control Project) sponsored by CECOM, we are investigating system architectures to support distributed applications a dynamic internet environment, an environment in which the properties of the communication and processing resources and their connectivity change dynamically. We employed a decentralized architecture in that all services and data are replicated and can be remotely accessly. We used a set of system tools, built on top of the operating system, to provide the required application services. These tools include a replication service, a name service, a resource monitor, resource manager, and database services. These tools comprise a distributed system. Its initial version has been deployed in the ADCCP testbeds to support reliable distributed communications and processing. The tools are currently being updated to provide additional capability. The system is based on the client/server model, with clients using remote procedure calls (RPCs) to communicate to servers. The RPCs are implemented based on the SUN Microsystems RPC protocol with some extensions to provide location transparency and reliable asynchronous communications. The replication service is used to replicate a server's actions at other copies of the server. Commands to servers are placed in a replication queue. Replication daemons periodically send these commands until they are received by the other servers. There are multiple replication queues and daemons. The name service is composed of a set of name servers, which are used to store configuration information about the system resources. This information can be changed dynamically, and changes are automatically replicated to the other name servers with the replication service. If the name server that a client has been using becomes unavailable, the system will automatically switch to another available name server. The resource monitor is used to obtain current status information about the system resources, including system services, hardware, and communications. The resource monitor is decentralized and hierarchical with resource monitors located throughout the internetwork distributed system. The resource monitors collect information about resources and store that information in a database. The resource monitors share information between each other in order to provide full system information at Westine [Page 8] Internet Monthly Report July 1987 each resource monitor. Clients request a generic service from the system. The specific server to use is specified by the resource manager. The resource manager uses dynamic system information from the resource monitor and service locations from the name server as inputs to a heuristic that determines the "best" server location for a requesting application. This provides location transparency and allows the system to make an informed allocation of system resources. The database services are based on the UNIFY (of UNIFY Corporation) relational database system. Unify was extended to support the client/server model of interaction and to support replication and consistency handling. The remote access interface implemented is a general SQL interface that operates using RPCs. Client side of the database uses the resource manager to locate and bind to a database server chosen from the replicas of that database. Ira Greenberg (Ira@spam.sri.com) and Joan Wrabetz (Wrabetz@spam.sri.com) UCL --- UCL continued with the TCP/IP performance tests using Van Jacobson's 4.3BSD TCP. Results are very promising, with unnecessary retransmissions reduced almost to none, and throughput performance up by a factor of 2.5 across SATNET paths. Work continues on IP throughput tests over SATNET with BBN to parallel the TCP work and gain a better understanding of TCP behaviour over these paths. Jon Crowcroft attended the End-to-End task force meeting. John Crowcroft (jon@CS.UCL.AC.UK) UNIVERSITY OF DELAWARE ---------------------- 1. Development continues on the Dissimilar Gateway Protocol (DGP). A technical note was distributed to the INENG and INARC Task Forces discussing several issues on IGP interactions, hierarchy rules and source routing. A briefing was presented at the INENG Task Force meeting. Westine [Page 9] Internet Monthly Report July 1987 2. The new fuzzware distribution for the NSFNET Backbone has been installed. Initial reports indicate that performance is even more improved over the version installed last month. During this month the packets delivered rose 50 percent to about 31 packets per second; however, the percentage of dropped packets fell by a factor of six to less than .09 percent. Observations on this and related issues were distributed to the tcp-ip mailing list. 3. Since the number of Internet nets has exceeded 250 and the number of NSFNET nets has exceeded 60, the routing tables in the M/A-COM Linkabit and Rice University fuzzball gateways have sloshed over the rim. Intricate changes to the fuzzball virtual-memory mapping tables have been completed and distributed, so that now the fuzzlings can handle at least 400 EGP nets and 100 IGP nets with the HELLO routing algorithm. 4. I attended the IAB meeting in San Fransisco and the INENG meeting in McLean, VA. With help from Hans-Werner Braun of the University of Michigan, I completed a paper on the NSFNET Backbone network for presentation at the SIGCOMM symposium in August. 5. Mike Minnich attacked several megabytes of performance data, harvested with help from Doug Elias of Cornell University, and produced a number of striking PostScript graphs detailing the performance of the NSFNET Backbone over the last nine months. Dave Mills (Mills@UDEL.EDU) Westine [Page 10] Internet Monthly Report July 1987 NSF NETWORKING -------------- NSF NETWORKING UCAR/BBN LABS NSF NETWORK SERVICE CENTER (NNSC) The RFCs describing the High-Level Entity Management System (HEMS) have been sent to the RFC Editor and Craig Partridge is currently writing an implementation of this system. The first issue of the NSF Network News was distributed and additional copies are available from the NNSC. Craig attended meetings of the Internet Engineering Task Force, the End-to-End Services Task Force, the SURANET User Support Staff Meeting and presented a paper at the Phoenix USENIX Conference. Karen Roubicek gave a presentation at the NSF-sponsored Supercomputing Institute held at the University of Georgia. By Karen Roubicek (roubicek@nnsc.nsf.net) NSFNET BACKBONE SITES CORNELL UNIVERSITY THEORY CENTER (Report not received) UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN Changes were made to the Cray CTSS IP/TCP to allow it to use the new Interprocess Communication Facilities have been implemented. The results of its usage is not noticeable in single stream throughput, but multiple concurrent transfers are faster. We also can now send packets up to 4k octets. Midnet has its lines to us installed, but DSU/CSUs have not yet arrived to make them usable. Would expect Midnet connection to NSFNET in August. Our ARPANET connection is moving again. The line is installed and should be connected by BBN in about 4 weeks. I have written a document with the following lofty goals: This document assumes that one is familiar with the workings of a non-connected simple IP network (e.g. a few 4.2 BSD systems on an Ethernet not connected to anywhere else). Appendix A contains remedial information to get one to this point. Its purpose is to get that person versed in the "oral tradition" of the Internet to the point that that net can be connected to the Internet with little danger to either. It is not a tutorial, it consists of pointers to other places, literature, and hints which are not normally documented. Since the Internet is a dynamic environment, changes to this document will be made regularly. The author Westine [Page 11] Internet Monthly Report July 1987 welcomes comments and suggestions. This is especially true of terms for the glossary (definitions are not necessary). It is available via anonymous FTP to uxc.cso.uiuc.edu and cd pub. There are two versions: hgi (in ditroff -me source) and hgi.txt (in text form). By Ed Krol (krol@uxc.cso.uiuc.edu) JOHN VON NEUMANN NATIONAL SUPERCOMPUTER CENTER This report is designed to inform the JVNNSC Consortium and JVNCNET network members as well as the Internet community of monthly status of the JVNCNET network. The data used on this report is collected using a number of techniques developed at JVNC, together with data from the JVNC operations group. The NSFNET data is obtained from the NSFNET NOC (Network Operations Center). Network Brief: JVNCNET has the configuration of a "tree". The 13 Consortium Universities, together with the 3 NRAC schools form the 16 node network. The gateways are 15 VAX's and 6 Ungermann Bass routers, connected by 9 T1 lines, 5 56kbps lines and 2 satellite links. Locally at JVNNSC (center of the JVNCNET network) we have a VAX8600 ("jvnca"), a VAX750 ("jvncb") used as routers and one UB dedicated router. "jvnca" is also (among other tasks) the "csc.org" primary name server and the JVNCNET primary monitoring machine. Network Statistics Data: The network monitoring statistics are performed from "jvnca", and are affected by its down time. In order to compensate for this we multiply (in our analysis below), the numbers with the percentage of down time of jvnca thus representing the worst case. We plan to overcome this situation with the next release of the JVNC monitoring package (see "Network Monitoring Project" below). The network is polled every 10 minutes with "icmp-echo" packets, and the analysis of the data is performed on this data. The data is processed using information on the "scheduled down time table" for the systems. Therefore, if a gateway is scheduled to be down, its information is not computed during that time, whether the gateway is operational or not. The special configuration of this network together with the fact that we use different subnets for the point to point links allows us to determine very accurately the reachability for the gateways. When we receive an alarm that one node/gateway is unreachable, we Westine [Page 12] Internet Monthly Report July 1987 proceed to determine whether the node is down or any other component of the the link is non functional. This information is not affected much by the routing. Table I, Gateways Reachability/Link Status ------------------------------------------------------------------ processed: Fri Jul 31 12:40:28 1987 node name meanttr maxttr meantbf sched_down avail perform (min) (min) (min) time (%) (%) (%) --------- ------- ------ ------- ---------- ------ -------- jvncb 19 60 2678 1 99 99 jvncc 42 524 1228 4 97 94 jvncd 51 474 1738 4 97 95 jvncf 48 145 4447 1 99 98 arizona 118 339 5036 0 97 97 brown 76 84 13690 0 99 99 colorado 52 159 4511 0 99 98 columbia 22 134 983 0 97 97 harvard 76 84 13670 0 99 99 ias 41 94 8207 0 99 99 mit 76 84 13670 0 99 99 njit 91 425 8052 1 98 98 nyu 122 2194 1752 0 93 93 penn_state 81 598 3667 0 97 97 princeton 134 389 10175 0 98 98 rutgers 71 169 8176 0 99 98 rochester 72 1229 1257 0 94 94 u-of-penn 218 475 13381 0 97 97 stevens 79 425 5812 0 98 98 umdnj 425 2931 5465 0 92 92 Total test time (min): 41240 ------------------------------------------------------------------ Note that the data is from July 1st through July 29th in order to process it before dead line. The Header definitions of table I are: nodename: the gateway or host on JVNC-NET network (128.121) meanttr: in minutes, the mean time to recover from the "down" state to the "up" state. Where "down" state is when the result of sending "icmp-echo" packets is no packets return, and the "up" state is when we receive packets back. Each test is performed 10 times, every 10 minutes, and averaged each time. Westine [Page 13] Internet Monthly Report July 1987 maxttr: in minutes, the maximum time to recover, from the "down" state to the "up" state (see above). meantbf: in minutes, the mean time between failures. sched_down time: in percent, is the time the gateways/hosts were "scheduled" to be down, respect to the total time of test. avail: in percent, is the time for which the gateways were available("up" state) respect to the total time of test (minus the scheduled down time). perform: in percent, is the figure of merit that considers the number of packets lost and the available time. Analysis: The data of Table I is a product of the gateways availability together with the lines/satellite status (this last including the communications equipment such as T1 muxes, CSU/DSU, etc). The following is broken down in gateways and links (lines and satellite). 1.- Gateways Availability: "jvnca" was up 97.43% of the time. Thus, the node that was less available was "umdnj" (97.43 * 92 % = 89.64%). In the average, the gateways were reachable 89.53% (worst case). Note that jvncc, jvncd and jvncf are the front ends for the two Cyber 205s. Table II, Gateways Availability ----------------------------------------------------------------- gateway school problems -------- ------ -------- jvnca JVNNSC hardware problems and power hits jvncb JVNNSC power hits jvnc-njit JVNNSC power hits, upgrade software njit-jvnc NJIT power shutdown njit-umdnj NJIT power shutdown njit-stevens NJIT power shutdown super PENN software upgrade to ULTRIX2.0 pusuper PRINCETON power outage psugate PENN STATE software crashed rochway ROCHESTER software crashed ----------------------------------------------------------------- Westine [Page 14] Internet Monthly Report July 1987 2.- Lines Status: Table III, Lines Status ------------------------------------------------------------------ from to type problem ----- -- ---- ------- JVNNSC MIT T1 at the CO in MASS. JVNNSC PENN STATE T1 mux at Penn State JVNNSC RUTGERS T1 AT&T down by mistake ------------------------------------------------------------------- 3.- Satellite Nodes Status: Table IV, Satellite nodes Status ------------------------------------------------------------------ from to type problem ----- -- ---- ------- JVNNSC ARIZONA satellite power outage JVNC JVNNSC ARIZONA satellite demodulator at Arizona JVNNSC COLORADO satellite power outage JVNC JVNNSC COLORADO satellite TransLAN software upgrade ------------------------------------------------------------------- 4.- Traffic: This data is collected from JVNCA, JVNCB and the JVNC fuzzball every 10 minutes. The data from COLO and JVAX (U. of Colorado and U. of Arizona) is estimated. The data from the jvnc-njit router and the dial-in/Tymnet terminal server are neglected. Table V, Traffic To JVNCnet (one direction) ----------------------------------------------------------------- local gateway packets comments ----------------------------------------------------------------- JVNCA 32,106,932 accurate JVNCB 11,358,646 low value due to missing data COLO 2,588,508 low estimate JVAX 1,571,268 low estimate FUZZBALL 18,816,315 accurate JVNC-NJIT 0 no data available TERM SERV 0 no data available ----------------------------------------------------------------- total....... 66,441,669 ========== The count obtained from the fuzzball is slightly distorted due to returned packets and excesive abuse from gateways ouside the JVNCnet system. Westine [Page 15] Internet Monthly Report July 1987 5.- PSN Status: Still no news from DCA when they will connect. Everything is ready. 6.- Comments: The area of the JVNNSC (including Princeton University, Rutgers, IAS, NJIT and others) was affected by power outages over the month specially affecting the VAXs. The network remained operational at least 89% of the time. This is a low estimate since we are combining the percentage of uptime for jvnca and the average percentage of reachability for all the gateways as two independent events. The routing has been unstable due to different situations, such as oscillations on the multiple-helloers with the NSFNET backbone's fuzzball, bad routing learned from certain Consortium networks, and hello packet size limitations on the gated code. Most of these cases were quickly determined and fixed. Projects: 1.- Network Monitoring: Status: On going The network monitoring package "netmon" has been designed to provide JVNCNET with the tools to quickly determine a network problem in our environment. This package has been in constant evolution as we have learned more on its limitations and the needs for monitoring. Today the so called "phase I" is completed and the program is fully operational. Still, "phase II" is on its way, to provide for new functions like the concept of "primary" and "secondary" monitoring center, at the same time "phase II" will allow for portability of the code, this will satisfy some of the JVNC Consortium's requests for copies of the program for their local use. Unfortunately we can only support the code for local use due to staff limitations. We are also looking very attentively at the efforts of the IETF's gateway monitoring group (chaired by Craig Partridge), and we'll follow their recommendations to enhance our monitoring facilities. 2.- Network Characterization: Status: On going The Network Characterization program is directed towards determining the parameters that characterize the JVNCNET network's diverse type of services. This effort will be utilized not only as a research subject but to find/predict network bottlenecks and problems before they are obvious to the end users. Westine [Page 16] Internet Monthly Report July 1987 This task started two weeks ago with the collection of data and the automation of the collection, and will continue with the study of the characteristics and patterns that distinguish each point to point lines whether T1, 56kpbs or satellite. The results will be available to the community. 3.- Traffic Analysis: Status: On going The traffic data is currently being collected from JVNCA, JVNCB, and the fuzzball automatically. This will be extended to the other VAXs on the same ethernet and the other routers on the network. At the same time the data will be compiled per campus. Miscellaneous: Brian Gould and Sergio Heker assisted to the Pittsburgh Supercomputer Center's Network meeting in Pittsburgh. Sergio Heker assisted to the Internet Engineering Task Force meeting in Washington, and participated of the short-term-routing committee. Information: JVNCNET NOC: "net@jvnca.csc.org" (JVNCNET Network Operations Center) JVNCNET manager: "heker@jvnca.csc.org" By Sergio Heker (heker@jvnca.csc.org) NATIONAL CENTER FOR ATMOSPHERIC RESEARCH AND UNIVERSITY SATELLITE NETWORK PROJECT The USAN backbone (128.116) will be will be connected to NAS at NASA Ames (128.161) via a Proteon router and 56kbps line by late summer. In early fall there will be a 56kbps line and a cisco or Proteon router connecting USAN with SESQUINET. The satellite links to the Institute for Naval Oceanography and the Naval Research Laboratory are still pending. The NCAR connection to the ARPANET PSN at the University of Colorado is still awaiting final resolution of some details. Westine [Page 17] Internet Monthly Report July 1987 First hop networks in back of the USAN gateways are now hello- announced to the NSFNET.RE fuzzball node at NCAR by gated. By Don Morris (morris@scdsw1.ucar.edu) PITTSBURGH SUPERCOMPUTING CENTER In July PSC-Gateway passed an average of more than a million packets per day through its two interfaces. The gated configuration has been kept up to date throughout the month. The machine only experienced a few hours of downtime during the month. Our Fuzzball also had high uptime. We brought up the most current software version on the 24th. We have had few line problems but at the end of the month began seeing numerous "buffer preemption" problems, the cause of which is being investigated. On July 15 another Proteon gateway was brought into service at Drexel University, connecting to the existing gateway at Penn. Our connection to SURAnet at the University of Maryland was down early in the month but has stabilized recently and appears to be performing well. A meeting was held for the networking contacts of our Academic Affiliate schools on July 9. We talked about the present status of the network and what we expect to happen in the near future. We discussed networking concepts, how we will connect to local campus IP nets, and provided them with sources of more information. By David O'Leary (oleary@morgul.psc.edu) SAN DIEGO SUPERCOMPUTER CENTER The newest version of SRI Multinet has been installed. This version adds support for GATED and name server. We are awaiting our GATED configuration information from the NOC before turning it on. At long last our IDNX node linking the Center to the University of California communication system is operational. It uses a single T1 link to the UCSD node, from ther the rest of the UC system is reached. The IDNX system carries across it a variety of (56k) links for SDSC including: SDSCnet to UCB, UCLA and UCI. DECnet to UCR. tcp/ip to UCSB, UCLA (to IMP 1 there) and UCB. Westine [Page 18] Internet Monthly Report July 1987 This latter link will be of most interest to this group, as it will also allow traffic with BARRnet, providing their direct path to NSFNET. We have also gotten the NASA supplied 56k (DECnet) link to SPAN at JPL operational. The T1 microwave link to the Salk Inst. (P4200 to P4200) has been tested. It is only awaiting software from Proteon to be brought online when it will carry both tcp/ip and DECnet traffic. By Paul Love (loveep@sds.sdsc.edu) NSFNET REGIONAL AFFILIATED & CONSORTIUM NETWORKS BARRNET (No report received) JVNCNET (Refer to JVNNSC backbone report) MERIT Merit member Michigan State University acquired and installed several more Suns this month, including one that is now their advertised backup domain name server, and have made progress toward upgrading all of the campus Suns to 4.3bsd-compatible software. They also upgraded their Vaxes to Version 3.0 of the Wollongong software. The Center for Information Technology (CITI) at the University of Michigan has been working on drivers for networking Macintoshes in a TCP/IP environment. They now have implemented Macintosh drivers for IP and UDP (currently deployed) and TCP (in test). ARP and ICMP are also supported. For name service they are using a domain name service client derived from "bind" or a Sun Yellow Pages client. At this time the Kinetic Appletalk/Ethernet gateway interface is deployed, while interfaces for integral Ethernet, serial IP, and the Kinetics SCSI/Ethernet gateway are in various stages of development. The upper layer applications already deployed are a TFTP client, Sun's XDR and RPC layers, and Sun's NFS (the latter as client only). At the University of Michigan site, considerable progress was made this month on the TCP/UDP interface to MTS (the Michigan Terminal System), one of Merit's major host systems. Hans-Werner Braun of the U-M staff attended meetings at PSCNET and of the Internet Engineering Task Force. By Christine Wendt (Christine_Wendt@UM.CC.UMich.EDU) MIDNET (No report received) Westine [Page 19] Internet Monthly Report July 1987 NORTHWESTNET Well, finally the last t has been crossed and the last i dotted. We have all the necessary NSF approvals, including that for the Boeing Computer Services (BCS) subcontract. BCS will be the facilities manager for NorthWestNet. Following our RFP for routers, BCS has selected Proteon boxes. The next step is to order the lines. By Helmut Golde (golde@june.cs.washington.edu) NYSERNET SUNY Fredonia was a new participant in NYSERNET in the month of July, its connection is currently 9.6kbits. Some additional 56kbit trunking was also added for redundancy and bandwidth in the month of July in the New York City Area. As of 1 August 1987, NYSERNET had the following topology: Clarkson Syracuse--+ | | | Rochester--------Cornell---------RPI---Albany | | | Buffalo...Fredonia | | | | | Binghamton | +-------- | ------StonyBrook | | | | | | | | CUNY--NYTEL/NSMAC--Columbia------NYU-+ | |\ | | /| | | | \ | NYNEX/S&T / | | | | \ BNL / | | | | \ / | | | | +-------------Rockefeller | | | | | | | +------------------------------+ | | | +-------------POLY-------------------+ The 4th Nysernet Technical meeting was held in New York City on the 15th of July. Participants discussed major technical issues and network applications that affect their campuses and users. A more rational approach to USENET "news" distribution was also agreed upon to remove more costly mechanisms. A presentation of the Simple Gateway Monitoring Protocol (SGMP) was presented at the IETF plenary session by myself. SGMP should become an RFC in the next months. Three independant implementations are scheduled for September delivery. Westine [Page 20] Internet Monthly Report July 1987 A root domain server was brought up on C.NYSER.NET and some initial testing and use was made by some NYSERNET members. Upon discussion at IETF about this issue, a new service offering with three new root domain servers is being put together (including C.NYSER.NET). Currently the root domain server is turned off. By Marty Schoffstall (schoff@nic.nyser.net) SDSCNET (Refer to SDSC backbone report) SESQUINET The complete initially proposed configuration of SesquiNet has now been completed. As a consequence the following campus networks are being served and being 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 Although we have not yet experienced what I would regard as heavy loads, it may still be significant that we have experienced no hardware or software failures of the cisco gateways once installed. A 56 kb/s line has been ordered to connect the NSFNET (fuzzball) gateway at NCAR to a fuzzball at Rice. This fuzzball's software will be controlled by the NSFNET fuzzball hackers. Since it has a direct connection to the ARPANET, this gateway will serve as an ARPANET-to-NSFNET gateway. Since it shares an Ethernet with the Rice SesquiNet gateway, it will also serve as a SesquiNet-to-NSFNET gateway. During August we will be experimenting with new cisco support for the Hello protocol; this will allow the SesquiNet's cisco gateway at Rice to dynamically advertise its campus networks to the ARPANET/NSFNET fuzzball gateway here. We have primitive tools in place to monitor network operation, and hope to be able to quantify our MTBF and MTTR for each of the networks during the month of August. As mentioned last month, we have a technical report that describes the network, and will be glad to send a copy to those interested. By Guy Almes (almes@rice.edu) Westine [Page 21] Internet Monthly Report July 1987 SURANET The SURANET User Support Staff Meeting was held at the University of Maryland on July 30 and 31. There were approximately 90 attendees. The meeting was highly successful. Videotapes of the meeting will be available. Requests for information about obtaining copies should be sent to Dr.Jack Hahn, HAHN@UMDC.UMD.EDU on, or after, August 24. Arrangements are essentially complete to add twelve additional sites to SURANET. The phone companies have promised installation by the first of September. All other components are already in place, or will be in place prior to this date. The twelve new sites will be: Louisiana State University University of Alabama, Seebeck Computer Center Georgetown University Gallaudet University College of William and Mary Johns Hopkins University Vanderbilt University George Mason University Catholic University of America University of Florida University of Virginia CEBAF (Continous Electron Beam Facility) By Jack Hahn (hahn%umdc.bitnet@wiscvm.wisc.edu) WESTNET I am writing this report as of July 20, 1987, so events described herein are up to date as of then. 1. Don Morris of NCAR does now have gate-d running on NCAR's Sun, with dynamic routing capability. As of now, we are still not being advertised to the Backbone, however IT APPEARS that this is imminent. Thank you to those of you who have assisted in this endeavor. 2. We have supplied five nominees to attend the Cornell TCP/IP workshop, and are looking forward to attending. 3. New Mexico Technet has agreed to cost share for line costs in New Mexico until the NSF FY'88 money arrives. A 56 kbps line to Wyoming might also become available (installed by Wiltel) also before the FY'88 money arrives. We are still exploring with AT&T whether they will cost share. It appears doubtful... Westine [Page 22] Internet Monthly Report July 1987 4. We have conducted an evaluation of IP Gateways. There were three respondents from the 16 or so vendors who were supplied RFP's. The hardware quoted was from Vitalink, cisco and Proteon. After a thorough evaluation, we made the preliminary selection of cisco. 5. We shall begin investigating vendors for CSU/DSU's next month. The cisco IP Routers come only with an RS-232 port. How droll... 6. During our NSF Summer Supercomputer Institute this week, Mike Moravan presented a workshop on networking. Of all the attenders, only about 1/2 had even heard of NSFNET. We need to do a better job of advertising on our campuses. By Pat Burns (pburns%CSUGREEN.BITNET) TASK FORCE REPORTS ------------------ APPLICATIONS - USER INTERFACE The first meeting of the task force took place June 25-26 at Stanford. The first day was devoted to presentations by the attendees, outlining their respective efforts with respect to user interfaces in/for/to distributed systems. Consistent with our charter, the majority of these efforts focused on multimedia communications. Several themes emerged: -- Contemporary window systems and user interface toolkits provide little support for multimedia information systems--in that they deal only with the traditional I/O devices: keyboard, mice, and display. Additional research is needed to determine how (the concepts underlying) these facilities should be extended to support voice and video, in particular. -- A multimedia workstation will almost certainly need additional (co-)processors to support voice and video--so as to avoid overloading the CPU. Multi-processor architectures are desirable for other reasons as well, as discussed below. -- For the foreseeable future, wide-area networks will be incapable of supporting the necessary (presentation) media (e.g. text and graphics, voice, and video) on a single transmission media. Rather, multiple transmission media--e.g. satellite links, terrestrial packet-switched networks, and analog telephone lines--should be employed wherever possible. However, clients should not have to specify a particular Westine [Page 23] Internet Monthly Report July 1987 transmission medium, but rather a "class of service", for each presentation medium; the system should dynamically choose the best transmission medium. -- Low-quality voice is not acceptable, particularly in the context of real-time teleconferencing. Based on current technology, this conclusion appears to rule out anything short of 24 Kbps (CVSD) encoding; 2400 bps LPC encoding, in particular, was unanimously rejected. This restriction in turn suggests that the only transmission medium that can provide reasonable coverage in the near term is the public telephone network. -- It is becoming technically and economically practical to couple contemporary workstations with "voice/phone servers". In such an environment, the voice server(s) provide call placement, answering machine services, voice store-and- forward, and speech synthesis. The workstation provides individualized directory services, database management (e.g. for annotated documents), and the user interface. Participants from Xerox, ISI, and the MIT Media Lab all reported on servers providing some subset of these functions. The remainder of the meeting addressed the last theme in detail. The goal was to specify, first, the partition of function between workstation and voice server so as to be workstation- and operating system-independent, then to specify a low-cost implementation that could be readily assembled within the next few months (and be compatible with a variety of workstations). Subsequently, other physical configurations, such as VME-based co-processor boards for Sun workstations, can be built to satisfy the same functional specification. Preliminary agreement was reached on the basic partition of function indicated in the itemized list above and on the following initial implementation: Primarily in order to permit experimentation with different voice boards (e.g. supporting different compression algorithms), and realizing that such boards will continue to be developed in greater quantity for IBM PC's, it was decided that the voice server should be a completely separate hardware unit from the workstation. The package of choice would consist of: - an IBM(-compatible) PC with - speech board(s), - hard disk for voice storage, - choice of Ethernet or serial line connections to workstation, - connection(s) to analog phone lines, PBX, or digital voice-and-data network, as appropriate. Westine [Page 24] Internet Monthly Report July 1987 The particular choices made with respect to the last two items should be invisible to both the user and to applications running on the workstation. A detailed strawman specification is currently under development. Topics to be addressed in the future include information interchange formats, reference models and implementation strategies for interactive software (in/for/to distributed environments), floor control strategies (for real-time teleconferences), and the user interface issues associated with the specification and monitoring of distributed computations. An additional distribution list, UI-INTEREST@VENERA.ISI.EDU, has been created for interested parties who are not members of the task force. Requests to be added to the list should be send to WESTINE@ISI.EDU. I will (re)post a detailed report of the first meeting to that list within a few days of this summary being issued. The next meeting of the task force is scheduled for October 14-15 in Cambridge, Ma. Keith Lantz (lantz@gregorio.stanford.edu) AUTONOMOUS NETWORKS Planning a meeting in Boston the second week of November. Deborah Estrin (estrin@OBERON.USC.EDU) END-TO-END SERVICES The task force held a one-day meeting using the multimedia teleconferencing facility between ISI and BBN. The 6 participants at ISI and 4 at BBN were linked with packet video and packet speech across the Wideband network. In addition, the group used BBN's Diamond within the mmconf teleconferencing executive to achieve the effect of an overhead projector. Although some technical problems arose, the meeting on the whole was highly successful, and the multimedia people at ISI and BBN were able to extract some useful lessons for improving the facility. VMTP Another release of the VMTP implementation within the BSD kernel is available from Dave Cheriton. VMTP has now been successfully moved into MACH, where it is showing very good performance. Cheriton has students working on various Westine [Page 25] Internet Monthly Report July 1987 experimental applications of VMTP, including gateway monitoring and the host name server/name resolver system. The meeting included an extended discussion of a number of design questions remaining in VMTP. The issues under discussion were: Entity naming, checksums, rate control, round-trip-time measurement algorithms, and security. MULTICASTING As we announced on the TCP-IP mailing list recently, a preliminary implementation of IP (host-group) multicasting (RFC-988) for 4.3BSD is available for distribution. It is temporarily being distributed by BBN; contact Craig Partridge. There is a new mailing list for people interested in VMTP and/or the HGM multicasting facility. To join the list, send a request to vmtp-ip-request@gregorio.stanford.edu. SUN PROTOCOLS Bill Nowicki of Sun gave a presentation on the current status of the various communication protocols which Sun has originated -- XDR, RPC, NFS, and Yellow Pages. An RFC on XDR has been published (RFC-1014), and the group gave Bill some feedback on a draft of an RPC RFC. He has also prepared an RFC draft on Version 2 of Sun NFS, and a preliminary draft on Version 3. PERFORMANCE Van Jacobson of UCB and Jon Crowcroft of UCL have been performing tests of 4.3+BSD TCP across SATNET, and Van has been further modifying his algorithms as a result. He should shortly have a TCP which will work effectively across a very wide range of Internet characteristics, including local Ethernets, congested gateways, and lossy long-delay satellite networks. We believe that the work which Van and Mike Karels have been doing represents a very important step forward in TCP implementations, and its wide adoption in the Internet would improve service for all. The task force discussed the mechanics of expediting the incorporation of this new TCP into vendor products which currently use 4.2BSD or 4.3BSD as a base. The essential idea is for Berkeley to freeze on an updated TCP, at an appropriate time which we hope is not too far away, and to make the frozen code available in the public domain. Van is pursuing this strategy. Westine [Page 26] Internet Monthly Report July 1987 The next meeting will be at MIT in October. Bob Braden INTERNET ARCHITECTURE The next meeting of the INARC will be as a workshop, with a format similar to the fond-remembered Internet Research Group Meetings. Topics will include the next-generation Internet protocol suite, with special emphasis on future internets of breathtaking speed and mindboggeling size, not to mention superior craftmanship in routing, congestion control and resource management. Noteworthy contributions may be published in the Computer Communications Review. The time and place of this landmark event will be announced shortly. Dave Mills (Mills@LOUIE.UDEL.EDU) INTERNET ENGINEERING 1) The task force met at the MITRE-Washington facilities on July 27-29th. The meeting was hosted by Anne Whitaker (head of the Protocol Engineering group) and David Wood (head of the MITRE-Washington Network Center). 2) The agenda for the meeting is given below: MONDAY, JULY 27th Morning: - Opening Remarks (Phill Gross, MITRE) - Working Groups convene in separate rooms - EGP3 (Mike Petry, UMD/Marianne Gardner, BBN) - Short-Term Routing (Chuck Hedrick, Rutgers) - Name Domain Planning (Doug Kingston, BRL) - Net Management/Gateway Monitoring (Craig Partridge, BBN/Lee LaBarre,MITRE) Afternoon: - Working Groups reconvene TUESDAY, JULY 28th Morning: - Working Groups reconvene - Short-Term Routing (Chuck Hedrick, Rutgers) - Name Domain Planning (Doug Kingston, BRL) - In parallel with Working Groups Westine [Page 27] Internet Monthly Report July 1987 - Discussion of ANSI Proposed Routing Protocols - Discussion of Long Term Routing Issues Afternoon: IETF Plenary - BBN Status Report (Bob Hinden/Marianne Gardner,BBN) - NSFNET Status Report (Doug Elias, Cornell/Hans-Werner Braun, UMich) - ARPANET Measurement Status Report (Phill Gross/Rob Coltun, MITRE) - Gateway Monitoring/Network Mgmt Working Group Report (Craig Partridge, BBN) WEDNESDAY, JULY 29th Morning: - A Plea from Vendors (Dave Crocker, TWG) - Comment on the International Internet (Peter Kirstein, UCL) - The Simple Gateway Monitoring Protocol (M. Schoffstall, RPI) - Working Group Reports and Discussion - EGP3 (Marianne Gardner, BBN) - Short-Term Routing (Chuck Hedrick, Rutgers) - Name Domain Planning (Doug Kingston, BRL) - Landmark Routing (Paul Tsuchiya, MITRE) Afternoon: - Round Trip Delay Estimation (Van Jacobson, LBL) - Comments on Fallback Routing (Dave Mills, UDel) - Dissimilar Gateway Protocol (Dave Mills, UDel/Mike Little, MACOM) 3) This was the first meeting with this particular format (ie, in which the first half of the meeting is devoted entirely to working groups) and, with some modification, may set the pattern for future meetings. The four main working groups (EGP, Name Domain Planning, Short Term Routing and Gateway Monitoring) had met previously and there were a total of 11 papers in various states of readiness. These include (working titles are approximate): Name Domain Planning- 3 papers - Domain Administrator's Guide - Domain Implementor's Guide - Plan Overview (Paul Mockapetris is also planning to issue an update to his RFCs in connection with this.) Westine [Page 28] Internet Monthly Report July 1987 EGP- 2 papers - EGP Version Handling - Proposed EGP Enhancements Gateway Monitoring- 4 papers - High-level Entity Monitoring System (HEMS) and 3 supporting documents Short-Term Routing- 2 papers - RIP Documentation - Proposed RIP Enhancements The recommendation of the EGP and Name Domain groups was that their work was finished after a final revision cycle and that their groups could be dissolved. New working groups are being considered for formation. Detailed working group reports from the meeting are in preparation and will be mailed to the IETF list when complete. Drafts of most of these documents are in the IETF directory at the NIC. For more information on obtaining copies, contact either the appropriate working group Chair (given above in the agenda) or Mike StJohns (stjohns@sri-nic.arpa). 4) The format of the IETF has evolved over the last few meetings, with the result that the meeting are now quite large. The April meeting in Boston (hosted by Bob Hinden) had almost 90 on its largest day and the July meeting had over 100. Meetings this large are, in fact, small conferences and our last two hosts deserve special thanks. In particular, Anne Whitaker deserves credit for a very well prepared meeting. We had 10 terminals for Internet access and even had guest login facilities (on a segregated machine) for working groups to edit documents locally (Thanks also go to Mike Louden, Mike Boring and William Malloy in helping to prepare these facilities). Phill Gross (gross@MITRE.GATEWAY.ORG) PRIVACY The IAB Privacy Task Force had a productive two-day meeting at NBS in Gathersburg, Maryland on 28 and 29 July. Attendees were: Dave Balenson, Curt Barker, Morrie Gasser, Steve Kent, John Linn, Dan Nessett, Rob Shirey, Miles Smid, and Steve Wilbur. Westine [Page 29] Internet Monthly Report July 1987 In the first day's meeting, we resolved several open issues left over from the previous meeting; an update to RFC989 will reflect these changes. We also discussed several comments received about the RFC, including one technical criticism which impacts the way in which authentication services are provided for multi-addressed mail. We reviewed the state of ongoing electronic mail privacy implementation activities at UCL and NBS, and agreed that it may be appropriate to inaugurate an implementors' group as an adjunct to the task force in order to address detailed implementation concerns. Copies of Dave Gomberg's access control model paper were distributed to the task force membership for review. The next meeting was set for 4-5 November at BBNCC. The second day was spent in joint session with the security SIG of the OSI Implementors' Workshop, which is trying to select approaches to message handling system security in the X.400 context. John Linn presented an overview of the Privacy Task Force's approach, and we listened to several presentations by others: Ted Humphreys of British Telecom about a MTA-based approach proposed to CCITT, Miles Smid of NBS about a key notarization mail system, Keith Howker of ICL about the ROSE project and office document architecture, and by Curt Barker of TIS about TIS's trusted mail system. 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. Westine [Page 30] Internet Monthly Report July 1987 TESTING AND EVALUATION No report received. Westine [Page 31]