================================================ n-2-1-002.01 President's Message Dear Internauts: As I write this, there is still a winter chill in the air, but the promise of Spring is evident in the daffodils poking their hesitant shoots skyward. Your Internet Society is preparing for the bloom of Spring, too. Trustee elections are in preparation and I want to thank Trustee Geoff Huston and his committee (Member Rob Blokzijl, Trustee Ira Fuchs, Trustee Tomaz Kalin, Pioneer Member Craig Partridge and Member Hide Tokuda) for their fine work preparing a slate of a dozen candidates. In addition, two successful petition candidates round out the slate to a total of 14 for 6 slots on the Board of Trustees. Information about the candidates is available on-line and will be sent with the balloting material later this Spring. IAB and IESG nominations are in progress and I want to thank Pioneer Member Jeff Case and his committee (Jack Brown, Pioneer Member Peter Ford, Pioneer Member Erik Huizer, Trustee Geoff Huston, Pioneer Member Stephen Kent, Pioneer Member Barry Leiner, Member Paul Mockapetris, Pioneer Member Craig Partridge, Pioneer Member Jon Postel, Jim Romaguera, and Pioneer Member Claudio Topolcic). They are working against an end of March IETF meeting deadline and their efforts are greatly appreciated. Internet Society information is now available on-line from a host at CNRI: ietf.cnri.reston.va.us. Please see the box elsewhere in this issue for a summary of what is available. INET '93 is coming up in August. General Chairman Eric Benhamou advises that there are still opportunities for corporate sponsorship (eric_benhamou@3mail.3com.com). Along these same lines, those of you whose companies or institutions are not yet organizational members of the Internet Society should give serious consideration this year. This year, for the first time, the Internet Society is sharing costs with the National Science Foundation, Department of Energy, Advanced Research Projects Agency and National Aeronautics and Space Administration in the support of the IETF Secretariat. Organizational sponsorship is essential to meet this commitment. Contact vcerf@cnri.reston.va.us for details. A new committee to explore potential Internet Society involvement in the use of the Internet in primary and secondary education is being organized. If you have interests in this area and ideas and energy to contribute, please contact Bruce Nelson (Bruce_Nelson@novell.com) or Kathy Rutkowski (kmr@cnri.reston.va.us). Finally, I want to remind all of you that the Internet Society is alert to opportunities to participate in, sponsor or otherwise encourage conferences and symposia relating to the Internet around the world. Proposals should be sent to Vice President Larry Landweber (lhl@cs.wisc.edu). The recent Computers, Freedom and Privacy Conference, IRTF Privacy and Security Research Group Workshop on Network and Distributed System Security, and Maryland Workshop on Very High Speed Networks are examples, as are the upcoming Joint European Networking Conference and National Net '93 Conference. I look forward greatly to seeing many of you in San Francisco at INET'93. Warmest regards, Vinton G. Cerf ==================================================== n-2-1-003.01 The Internet Peace Corps by A.M.Rutkowski Some thirty years ago, U.S. President John F. Kennedy rallied a generation to a new form of international public service at the grass roots. The Peace Corps was created to motivate young and old to provide their own skills, creativity and energies to make a difference in working with other peoples around the world. In an ultimate form of human leveraging, the Peace Corps measurably improved the condition of countless villages and millions of people throughout the Third World. In the process, it also changed the lives of those who volunteered, and established a proud new standard of assistance for a nation and the world. One of the more remarkable attributes of the Internet is the ability for this enormous network of networks to bring fundamental change and empowerment to people throughout the world. One of the more heartening attributes of the Internet community is the number of individuals and institutions that have effected a kind of collective virtual Internet Peace Corps. Many pages of the Internet Society News have been devoted to their remarkable efforts and achievements. Some have literally devoted their lives to bringing Internet connectivity to technologically developing countries of the world. In addition, the Internet Society has begun working with international health and disaster relief agencies. This activity began as existing NGOs (Non-Governmental Organizations) of many kinds began to use the technology to improve the connectivity of very low budget operations around the world. The Institute for Global Communications was created out of this need. Activists in many diverse organizations with international missions - from the UNDP to the U.S. NSF to France's Orstom organization - then began to fund Internet connectivity and information infrastructure initiatives as express objectives; bypassing the tired and trite activities of the more traditional organizations that produced more paper than infrastructure. The process unleashed considerable technical and operational creativity to match prevailing local conditions. The networks had to be cheap, easily maintained, robust, and often optimized around enormously expensive and poorly performing public telecom capabilities. Far flung networks based on Unix UUCP or on DOS PCs like FidoNet were born and given gateways to the Internet. As the expertise grew and scarce funds were marshaled, circuits were shared, and full Internet connectivity was obtained. Governments around the world are now focusing on their national or regional information infrastructures, and turning the swords of the Cold War into plowshares. They are also facing many serious economic constraints. The current efforts of the new Clinton-Gore Administration are particularly notable examples. In these efforts, the enormous value of a more formal Internet Peace Corps initiative that would bring Internet connectivity and information resources to Third World locations is worth considering. Even more than the original Corps program, it leverages existing resources, people and technologies. It also captures the imagination and establishes a spirit of global change and cooperation in ways never before possible. ==================================================== n-2-1.011.10 A Latin American and Caribbean Update by Daniel Pimienta*, REGIONAL The process of regional integration has been slowly progressing in the last year; much faster improvements have been made in terms of sub-regions. If the very uneven level of institutionalization of the various national initiatives, and the weaknesses in the coordination between the different actors of the scientific information circuits (producers, consumers, carriers) makes the path long and difficult toward a real regional network integration, the level of mutual knowledge and cooperation between the people involved in networking made a spectacular jump in 1992. The Second Regional Networking Meeting was held last November 1992 in Guadalajara (Mexico). It was characterized by an almost exhaustive representation of national network initiatives and a more reduced international presence (OAS, UNDP, NSF, and SELA as observers), as compared to the first meeting in Rio de Janeiro. The agreed final text shows a lot of progress toward the regional perspective, and definition of the key strategical objectives for the region. The next meeting, scheduled to be held in Venezuela, in October 1993, should show the maturation of sub-regional and regional efforts. The Latin American Networking School of Merida (Venezuela) confirmed the promises demonstrated during the previous Rio de Janeiro Workshop. The regional ability to provide a comprehensive and high quality technical training program has been proven. One possible area of improvement could be in completing the curriculum with an introductory perspective which could sensitize the future network specialists on users, services and the social impact of networks. The REDALC regional project feasibility study (EEC funding, Union Latina, Unesco and Acal partners) was completed by September 1992. Follow-on actions are being designed and will be soon negotiated with regional counterparts. The REDALC team is also putting together, in collaboration with various partners, a plan for the creation of the Haitian network before the end of 1993. NATIONAL Practically the only country left out of reach from the Internet is Haiti. The countries with older experience are starting to get organized for network applications, while the younger ones are trying to consolidate their user bases. The Peruvian net continues to demonstrate its model value in bringing the focus toward user support and services, and leading regional actions. In March 1993, the Networking School will offer regional services in Lima. The event celebrates the first birthday of the Peruvian network and is organized together with various sub-regional or national activities. Two important improvements in terms of international connectivity have been made in Costa Rica and Ecuador, where direct Internet connections have been set-up. Both use 64kbps point to point satellite (PAS-1) connections to the Sprint PanamSat teleport in Homestead, FL and optical fiber from Sprint to NSF, Washington DC. Both are offered to be shared by the different national initiatives, respectively by Universidad de Costa Rica and Banco del Pacifico (in the last case, it is a subset of a business bandwidth). Venezuela is keeping a high user growth rate, and improving its network architecture with multi protocol support. The Cuba network (CENIAI) growth in users and services remains impressive: more than 2000 users are now connected. CENIAI has interesting assets to show in the upcoming INET93: the synergistic relation between telecom authorities and research networks, the special quality of user support, and the focus in applications. The Internet Society has to work to make possible the representation of CENIAI in San Francisco. * Redalc Projects Manager ================================================ n-2-1-011.25 Description of the Research Network Initiative in Costa Rica by Guy F. de Teramond* Two years after establishing the first Bitnet node in the Central American region, the University of Costa Rica interconnected last January 26, twelve nodes to the Internet. Major activity at present is aimed at establishing CRNet, a digital high speed national internet backbone, linking major institutions in the country. We provide a description of the research networking activities in Costa Rica, initiated three years ago. Completion of this program will enable the regional scientific community to access remote advanced computing and information systems, as well as offering information on important regional aspects such as biodiversity and tropical medicine. I. Establishment of a Bitnet Node in Costa Rica A Bitnet node is operational at the University of Costa Rica (UCR) Centro de Informatica. The node was connected on November 8th, 1990. A plan for the installation of a Bitnet node at UCR and the electronic connection of Central American Scientists was originally presented as a formal project to the Space Conference of the Americas in March 1990 (1). With the installation of a 3708 protocol converter and a PBX at ICE (Instituto Costarricense de Electricidad), and a second 3708 protocol converter at one of the RACSA (Radiografica Costarricense S.A) X.25 nodes, the system has convenient access from any PC, through the normal telephone lines, or through a data transmission line from a RACSA node, in the country or in the Central American region. The node at the University of Costa Rica is connected to the node FAUVAX at Florida Atlantic University (FAU) with a dedicated 19.2 Kbps digital link of the satellite PAS-1 from Panamsat. The receiving station of PAS-1 is located at Homestead, Florida. The number of registered users of the Bitnet node (UCRVM2) at the University of Costa Rica has been growing steadily since November 1990, when the node was interconnected. There is at present some 1500 users, most of them faculty, and students from the University of Costa Rica. Some 400 users are outside host members from some 30 institutions and specific research projects. A Bitnet NJE (Network Job Entry) link has been established with Panama over a 9.6 anological link. The new node UTPVM1 is operational since October 1992. There also exists a UNDP funded project at the Fundacion Nahual - Proyecto Huracan - which gives useful electronic mail services within Central America, also using RACSA X.25 packet switching network, with a dialup connection of the main node Huracan with a UUCP gateway in the US. II. Interconnection to the Internet The IBM Corporation has given the network project at the University of Costa Rica a grant which includes a RISC 6000/530 multiprotocol machine, that is interconnected to the 4381 computers at UCR. The RISC serves as hardware platform for the root name server (top domain level of the Costa Rican Internet). Bitnet traffic will be combined with native IP traffic by encapsulating Bitnet RSCS packets into IP datagrams using the VMNET transport software developed at Princeton University in the 4381 VM machine, locally migrating to BITNET II. A 64 Kbps satellite link has been established to carry the IP traffic to the US. On March 1, the 19.2 NJE Bitnet link to FAUVAX will be disconnected permanently. The UCR has already interconnected a dozen nodes to the Internet. Other nodes at various faculties and laboratories will be soon interconnected as a router- based fiber local backbone will span across the university campus. The Instituto Tecnologico de Cartago has recently acquired 12 high performance DEC STATION 5000 establishing also a local internet. III. Establishment of a National TCP/IP Backbone In parallel with the Interconnection with the US Internet, the National Research Network of Costa Rica (CRNet) has been established. This pioneer project, sponsored by the Ministry of Science and Technology, for the establishment of an internet backbone within Costa Rica is based on a proposal to the Agency for International Development (2). CRNet is a digital network that will provide high speed local interconnectivity between scientists at universities, research laboratories, industries with a technological component and other national institutions. CRNet also provides high level interconnectivity with the US Internet using RACSA and Panamsat teleport facilities located respectively in San Jose and Homestead (Florida), and a 64 Kbps digital link from the satellite PAS-1. The circuit includes a fiber link from Sprint corporation between Homestead and the NSFNet gateway in Washington D.C. sponsored by the National Science Foundation (NSF). The backbone will initially interconnect the University of Costa Rica (UCR), the Instituto Tecnologico de Costa Rica (ITCR), the Centro Agronomico Tropical de Investigacion y Ensenanza (CATIE), the Omar Dengo Foundation (FOD), the Escuela Agronomica Regional del Tropico Humedo (EARTH), the Universidad Estatal a Distancia (UNED), the Sistema Nacional de Informacion Cientifica y Tecnologica (SINICIT), The Universidad Nacional (UNA), the Instituto Nacional de Biodiversidad (INBIO), the Instituto Centroamericano de Administracion de Empresas (INCAE), the Fundacion NAHUAL (Proyecto Huracan), the Earth Council, and the Instituto Interamericano de Ciencias Agricolas (IICA). The Instituto Costarricense de Electricidad (ICE) has recently installed 140 Mbps optical links between major cities in Costa Rica. A 2 Mbps (multiplexed out of ICE 140 Mbps intra city links) is administered by Radiografica Costarricense S.A. This is referred as the RACSALINK NETWORK, the logical circuit and physical link for the TCP/IP Internet Costa Rican Backbone. The 2 Mbps RACSA links (speed E1) are multiplexed in 30 56/64 Kbps channels. Central RACSALINK network nodes are located at ICE/RACSA facilities at San Jose, San Pedro, Cartago, Alajuela, Heredia, and Limon. RACSALINK Network is a distributed architecture state of the art multi-trunk multi-node fiber optics network. The system is based on CODEX 6281 network processors, located at central ICE/RACSA nodes. At the network level, CRNet backbone is based on high performance CISCO routers (AGS/4 and MGS/4) located at RACSA/ICE nodes to route IP traffic and to serve as Point-of-Presence (POP) to interconnect individual institutions. A project sponsored by the Organization of American States (OAS) will permit the interconnection of the Universidad Nacional de Ingenieria in Nicaragua to the CRNet Internet gateway. It is expected that this connection will be the starting point of a Central American internet backbone. Initially, IP traffic to Central America will be carried over 9.6 analogical links. IV. Conclusions An important networking activity prompted by the recent interconnection to major worldwide research networks has taken place in the country. This activity is now centered in the establishment of local internet nodes interconnected by a local TCP/IP fiber backbone. Without access to the world scientific potential, scientists in the country will remain isolated, frustrated and unproductive. This project represents an opportunity to learn and implement scientific communication tools as well as giving a major impact to the whole education system and country economy. Acknowledgements Funding for the interconnection of Costa Rica to major research networks comes from a BID-CONICIT grant, the University of Costa Rica, the IBM Corporation, the Agency for International Development (AID), and the Organization of American States (OAS). In addition to the decisive support from the University of Costa Rica in personnel, equipment and financial resources, important support has been received form the Ministry of Science and Technology, the University of Wisconsin-Madison, the National Science Foundation (NSF), Racsa, Sprint, Panamsat and cisco Corporation. Additional support has been received from NASA Goddard Space Flight Center and Florida State University. References (1) M. Cerdas, G.F. de Teramond and C. Gutierrez, An International Electronic Connection for Central American Scientists, Presented at the Space Conference of the Americas, San Jose, March 12-16, 1990. Conference Proceedings, Vol. 2 p. 680 (San Jose, 1991). (2) G.F. de Teramond, C. Gutierrez, E. Mata, R. Oreamuno, L. H. Landweber, R.D. Bremel, Establishment of an Internet Backbone Within Costa Rica, Proposal to the Agency for International Development, San Jose, 1991. *Director, R&D Unit in Information Technologies and Networks, Universidad de Costa Rica, San Jose, Costa Rica ================================================ n-2-1-012.05 Ebone provides Megabit IP backbone At the Ebone consortium meeting on February 3rd in Luxembourg, the partners finalized the budget, decided on an upgrade of the backbone, set up the organization to operate Ebone in 1993, and confirmed Ebone's long term strategy. 21 partners have confirmed their commitment to the budget which covers a backbone connecting six backbone sites, nine additional access lines and three links to the US. Ebone is thus used directly for their international IP traffic by partners in virtually all Western European countries and indirectly by several other countries linked to the Ebone partners. In addition, Ebone will be used for pilot CLNS traffic. At the meeting, a plan to upgrade part of the backbone was decided. This means 1.5 Mbps from Stockholm to Washington and to Amsterdam, 1.5 Mbps from Amsterdam to Geneva, 1 Mbps from Geneva to the US and 2 Mbps from Geneva to Paris, and 1.5 Mbps from Paris to Washington. The US links are provided in cooperation with NSFnet and other US IP providers. In addition, 256 kbps links from London to Paris and to Stockholm are being considered for upgrade and the inclusion of a London to Washington link is being investigated. A new backbone site in Bonn has been decided and one in Vienna is being investigated. With this much needed upgrade Ebone can more confidently transport traffic from the rapidly growing number of IP hosts in the partners' networks. Kees Neggers was re-elected as Ebone chairman and an executive committee consisting of Dennis Jennings, Phil Jones, Christian Michau, Dave Morton, Kees Neggers, Peter Rastl and Peter Villemoes was appointed. Funded staff which will assist the executive committee and operate the network includes Frode Greisen as (part time) general manager, Bernhard Stockman as (part time) chairman of the Ebone action team which is planning Ebone development, and Peter Lothberg as Ebone operations center manager assisted by full time staff at KTH in Stockholm and part time staff at each of the six Ebone backbone sites. RARE is providing the secretariat support for the consortium. The meeting also confirmed Ebone's long term strategy to concentrate in the future on providing a neutral interconnect for all networks, while it is assumed that provision of backbone services will be offered by one or more (competing) providers in the longer run. Until such offers are forthcoming, Ebone will take care of its partners' needs in this area too. ================================================ n-2-1-012.07 EuropaNET PTT Telecom & COSINE announce the launch of EuropaNET, a pan-European multi-protocol backbone data network service The Eureka development project COSINE, funded by the countries of both the European Community and EFTA, has as its central objective the provision of a pan-European telecommunications infrastructure for academic, industrial and governmental researchers. Following completion of a public invitation to tender by RARE (Reseaux Associes pour la Recherche Europeenne) on behalf of COSINE, a multi-million ECU contract has been awarded to PTT Telecom of the Netherlands to provide a backbone data network and associated services supporting a range of communications protocols currently used in the research community. The three year contract includes performance and availability guarantees; these in themselves represent a major step forward in the procurement of international telecommunications services within Europe. The network, to be known as EuropaNET, can be accessed directly at national points of presence within all the COSINE member countries and already currently offers a range of access speeds up to 2 Mbit/sec. Independent performance analysis which has just been completed indicates that the network can efficiently provide for transmission of both TCP/IP and X.25 packets, the standard communications protocols currently used in the European research community. It is expected that all European national research networks will connect to EuropaNET, which will also have intercontinental connections to similar networks worldwide. EuropaNET has replaced IXI, the backbone network which had provided X.25 services to the research community across Europe for the past two years. Chairman of the COSINE Policy Group and Advisor to the Director of the Telematics Division of the European Commission's DG XIII, Horst Huenke said, "EuropaNET offers the European research community the opportunity to pool their resources into a single backbone network, thus obtaining significant economies of scale. The multi-protocol capabilities of the network should render irrelevant the arguments about rival technologies which have tended to distract the European research community from making progress in this field." In addition to the multi-protocol backbone service offering both X.25 and TCP/IP interfaces, EuropaNET also provides the ability to interwork between different access protocols and a development path to a range of future interface standards. It is based on an extensive investment by PTT Telecom in lines and infrastructure, with switching technology being provided by RC Electronics. In order to encourage national research networks to take advantage of the high performance 2 Mbit/sec. access, the Commission of the European Community is making available enabling funding to assist networks with their subscriptions. Two networks have already connected at this speed and a further four, including three interfacing via TCP/IP, are expected to connect at this speed in the next few months. Access to EuropaNET is being extended to Eastern Europe by a separate contract between the European Commission and PTT Telecom as part of the Commission's support programme for Eastern Europe. As a result, researchers in Poland, Hungary, the Czech and Slovak republics, Bulgaria and Romania will be able to communicate with researchers world-wide using EuropaNET. It is expected that the Eastern European countries will be connected to EuropaNET during the course of this year. Unisource Business Networks has taken over the operation of the network from PTT Telecom and will continue its management from their International Network Management Center in The Hague. Unisource also sees it as a major asset and opportunity to position and use this network for other advanced European Network requirements and projects in other areas where the EC has a vested interest. How and when the services will be integrated into the Unisource Business Networks service portfolio will be decided during this year. ================================================ n-2-1-012.42 Evolution of the GARR network by Stefano Trumpy* GARR is the acronym for Harmonisation Group for Research Networks created in 1988 operating under the Ministry of the University and of Scientific and Technological Research (MURST) in Italy. GARR is also the name of the Italian Research Network which is currently conducted by the founder organisations: three public research nation-wide Institutions i.e. CNR (National Council for Research), ENEA (National Council for Research on Nuclear and Alternative Energies Sources), INFN (National High Energy Physics Institute) and by three consortia offering computing resources to Italian universities, i.e. CINECA, CILEA, and Tecnopolis-CSATA. The aim of GARR is to establish and operate a backbone interconnecting the Italian research and academic networks and to co-ordinate connections to international networks. All computers on GARR will use Internet-style domain addresses. The top-level domain is .IT (Italy) and the user Internet addresses are in the form user@domain.it. GARR will continue to maintain connections to the major research networks, including EASInet, Internet, BITNET/EARN, EUnet, Fidonet, HEPnet, Europa Net, EBONE and other networking initiatives. The participation of Italy to the COSINE project was co-ordinated by GARR and particular attention is given to the follow on activities of the project i.e. the operational Unit and the Europa-net infrastructure. The backbone of the network provides four TDM channels over 2 Mbps lines, carrying IP, DECnet, SNA and X.25 traffic. The backbone was initially built up by the original seven primary sites located in Milano (CILEA), Bologna (CINECA and CNAF-INFN), Pisa (CNUCE-CNR), Roma (ENEA and NIC- INFN) and Bari (CSATA). Subsequently MURST funded a project to connect the over fifty universities in Italy ; the major ones to be connected as extensions of the backbone (2 Mbps), while the others to be attached with 64kbps lines to the primary sites. Also GARR is adopting a new public service providing bridging betweek ethernet LANs at 2Mbps (CLAN). The present configuration of the backbone is shown in the following picture. The planned extension under grant of MURST is the following. The GARR network has been constituted to serve primarily the following users of all institution reporting to MURST: -Universities -Consortia offering computer services to universities (CILEA, CINECA, etc.) -Astronomic and Astrophysics Observatories e Vesuviani -National Research Council (CNR) -National High Energy Physics Institute (INFN) -Italian Space Agency (ASI) and -other public funded research institutions like ENEA (National Council for Research on Nuclear and Alternative Energies Sources) and the Consortium Tecnopolis CSATA etc. The network services are also accessible by research departments of private initiatives which have cooperations and common projects with public funded research environment. The utilisation of the network is allowed for the activities connected to the development of research programs, higher education, to the diffusion of scientific information and to the support and management of the research programs. The utilisation of the GARR network is not allowed for - commercial activities; - exchange of information not pertinent to the scientific research or higher education; -not permitted access to the facilities connected to the network for a usage that provokes waste of resources, infringement to the privacy, etc. The GARR Network Information Service (GARR-NIS) is located at CNUCE-CNR in Pisa, which currently provides it with all the necessary resources: hardware, software, financing, personnel, etc. The GARR-NIS is running the IT top-level domain name server and the c=IT X.500 DSA. GARR NIS is the reference-point for various registration procedures (for organisations, networks, domains, persons, etc.) required to access to the GARR network and to international networks. GARR NIS main functions are: to be the reference-point for various registration procedures (for organisations, networks, domains, persons, etc.) required to access to the GARR network and to international networks. to define procedures and registration templates, taking into account the information requested by other European and American NISes (GSI-NIC, NNSC, RIPE-NCC, EASI- NIS, etc. ), in accordance with the Guidelines established by GARR and collaborating with the international networks managers. to acquire network information and to mantain databases and directory services in order to make such information accessible via the network. to provide the Internet DNS and X.500 DSA services at the national root level, and to coordinate their distributed management at the national level. to publicise, duplicate or improve the accessibility (through directory and/or information discovery systems) to useful network information provided by other NISes or information-servers. to carry out promotional activities spreading information on the GARR network and NIS: periodic reports, announcements, news about workshops and seminars, etc. GARR-NIS is reachable at the following address: via Internet at INFO@NIS.GARR.IT and via X.400 at c=it; admd=garr; prmd=garr; o=nis; s=info [I am sending via fax to ISOC a copy including two pictures] *CNUCE - ISTITUTO DEL CNR PISA ================================================ n-2-1-012.55 New Developments in Hungarnet by Istvan Tetenyi, E-mail in Hungary The coverage with e-mail of Hungarian users has significantly improved in the last couple of months. Today, e-mail is the most popular service which is provided for the R&D community. Users are e-mail "aware" as the first nation wide service called ELLA was introduced as early as 1988. This rapid growth was foreseen and therefore the technical committee of Information Infrastructure Programme devised a technical guideline in late 1991, on how to improve the service. The dilemma between choosing only one mail protocol and the diversity of options was very serious. Mail protocols like SMTP, BSMTP, Mail-11, UUCP and the home grown ELLA protocol were all in use. The transport protocols were also very rich: X.25, TCP/IP, NJE, DECNET and matters like RFC 822 addressing had to be introduced. The solution finally chosen was based on the brilliant package of M. Maddison, the MX mail exchanger for VMS. It allowed to integrate fully the ELLA protocol with all the other mail systems. One very significant feature was the full support of SMTP over X.25. The MX package had all those features which provided a sound basis for growth without protocolar constrains. System administrators found it easy to manage MX and sites with VMS generally opted for it and this will be the general solution for the coming years. The HBONE project The aim of this project is to establish a TCP/IP backbone for the Hungarian research and education community. The primary role of HBONE is to provide a private data network for TCP/IP networking. To interconnect the backbone routers initially, 64kbit/sec speed leased lines will be used. The X.25 service of the PTT will also be used in some cases as primary and in general as backup solution. It is expected that by mid 1993 a star shaped network with six leaves around Budapest could be finalised. Router suppliers were asked for proposals in 1992, and not surprisingly, cisco was given the option to provide the first ten routers for the R&D community. HBONE will be connected to the FDDI backbone of three university campuses in Budapest and also the FDDI ring in the largest county city, Debrecen. International connectivity is provided by two 64 kbit/s leased links to EBONE via neighbouring Austria. The European Multiprotocol Backbone (Europenet) provides another 64kbit/s X.25 link which will be used also for IP traffic. The network management of the HBONE is undecided as yet. Initially, a cooperative management was adopted until the number of routers are below a given limit. Last year, a Charta of acceptable usage policy was developed and accepted by backbone managers and end users. There are concerns about the available bandwidth. The number of active users in university campuses has grown with the completion of LAN/MAN-s. Sites with 1000 users are not rare anymore. The bandwidth per capita (a quantitative measure similar to GNP per capita) is very low in Hungary, therefore the preparations for introducing higher speed on HBONE links have been started, on some of the national links 2Mbps and on international links 128 kbps will be introduced in the next year. Great emphasis is on client/server type applications. The usage of batch oriented services like NJE is preferred. This is the reason why we are expecting the need for rapidly growing speeds. Sites without real TCP/IP connectivity are also cared for. Mail service, gopher searches, and News reading are publicly available for users with simple X.29 access. ================================================ n-2-1-012.56 Romania by Stephen Ruth* ruth@gmuvax.gmu.edu In less than two months, six major nodes have begun BITNET/ EARN transmissions. The Institute for Atomic Physics in Bucharest (ROIFA), led by Dr. Ghorghi Pascovici, has already registered five hundred users, aiming for a thousand, and is in contact with countries all over the world, aided by a grant from the Mellon Foundation. Currently they are the busiest node but the country's largest university, Politechnic Institute, (ROIPB) will soon be connecting several thousand users, according to Dr. Paul Christea and Dr. Nino Popovici, who are leading the effort there. Part of a Mellon grant is funding a group of highly skilled graduate students at IPB who call themselves the BEST team. They are doing some of the training and working with their professors to become skilled networkers. Current nodes include: ROEARN, routing-only destination, in Bucharest has about one hundred users; Eugen Staicut : ESTAICUT@ROEARN, and George Macri : GMACRI@ROEARN. ROIFA (Institute for Atomic Physics) growing to over one thousand users; Serban Constantinescu SERBAN@ROIFA ROIPB (Politechnic Institute) up and running now at full capacity due to the recent COCOM approval, with a total potential of thousands of users; ROBCU (Central Univ. Library) recently connected, mainly a D.B. server, but with about a hundred users already ROUTT (Technical University of Timisoara) with a large potential but only a few terminals ROIMAR (Institute of Mathematics) with potential of connecting the Romanian Academy of Science to a powerful DEC-based system George Gussi,director of IMAR Gussi@ROEARN.BITNET or GUSSI%ROEARN@VM.UNIVIE.AC.AT The Academy : ACADROM@ROEARN.BITNET A ring of FIDONET stations was established around Bucharest in November , enabling dozens of new nodes to be established in a matter of weeks. The FIDO stations, first in Romania, have complimented the more robust network connections. FidoMaven Bob Barad's outstanding skills in the setup made this a reality. Romania is gradually building up its infrastructure in terms of people and equipment as part of a long term plan. Romanians are also connecting in record numbers to the Internet and EARN because they realize that every new user helps to develop a greater opportunity to spread the benefits of networking. They are also investing in spreading network benefits, even barely realized ones, to Brazov, Iasi, Timosoara and other cities outside of Bucharest. They are poised for major progress but will need a lot of energy and sharing of resources internally and more help from the private-sector, governments, technology associations and interested Internet Society members every step of the way. A test of the country's resolve will be the degree to which humanists, liberal arts specialties and other disciplines that are often the last to be allowed to use network services are empowered to participate. Dr. Mahi Dragonescu, the president of the Romanian Academy of Sciences recently said that the Romanian plan is to have at least one humanist connected to the network for each engineer. If this trajectory is achieved Romania's explosive entry into networking will be sustained and the country will become a model for eastern Europe and beyond. * Professor, George Mason University ================================================ n-2-1-012.70.3 Russia (GlasNet) by Anatoly Voronov GlasNet, the Russian network for democratic communications, is about to reach the one thousand users threshold in a couple of months. The Russian economic situation, with hyperinflation and lack of political stability, forces us to be extremely inventive and astute in order to make ends meet. Yet, positive changes are taking place. A new packet switching service, Infotel, has been settled in Moscow, offering enhanced facilities of access. Actually, an intercity call is almost the only way of access GlasNet host from other cities. Starting April, it will be possible to access GlasNet by local PADs in Ekaterinburg, Novosibirsk, Rostov-on-Don. A PAD in St. Petersburg is up and running. The service is being planned to expand to Novgorod, Barnaul, Krasnoyarsk, Voronezh, Vladivostok. GlasNet, with its reasonable rates and policy which encourages private persons to use telecommunications, attracts the interest of academic and research workers, offering them an opportunity of direct and independent connection with their colleagues in the USA and other Western countries. GlasNet's heavy users are Protein Research Institute in Puschino (Moscow region), Lebedev Institute of Physics (Moscow), Vernadsky Institute of Geochemistry (Moscow). The traditional bottleneck is the bandwidth to the West. The dialup connection to the IGC (Institute for Global Communications) host in San Francisco, California, is growing more and more expensive. In January, rates for international calls increased four times. In this context, GlasNet is looking forward to get an access to a new 64 kbit/s channel granted by the Soros Foundation, to foster better information facilities for Russian academic workers. The appeal for used modems and other hardware, placed in the previous ISOC News issue, has worked: David Lisbona sent us from Israel a modem and a laptop, to give them "second birth" in Russia. Note to Tony, Steve G. handed this "submission" to me at the NORDUnet conference while we were in Helsinki. He said he didn't know where you wanted to put this in the News, but trusts us to place it appropriately. ================================================ n-2-1-012.70.4 Soros Generosity Assists Russian Connectivity by Steve Goldstein In December 1992, George Soros, the Hungarian-born U.S. financier and philanthropist, announced the formation of the International Science Foundation (ISF) for the Former Soviet Union (FSU), and his intent to donate US $100 million over two years through the ISF to help stabilize the practice of science and relaxed scholarly activity in the countries which have emerged from the FSU. In a series of meetings among advisors to the ISF, a strong consensus and clear message emerged: "Internet connectivity is vital to the well-being of scientists in the FSU". Mr. Soros received the message favorably, and now telecommunications is an integral activity of the ISF program. The National Science Foundation (NSF) assists the ISF by providing advice and access to selected NSF networking infrastructure support activities. The immediate focus will be assisting ISF's work with the Russian science community to implement one or more 64kbps external (international) links. Similar activities with other FSU countries is also being pursued. The next steps will be to help with the improvement of in-country connectivity in the FSU countries. Also, ISF will consider supporting activities to train service providers to provide assistance to end users and to help prospective end-users gain access and learn how to use the networks. In all instances, ISF will seek cooperation with existing service providers and work with other international assistance activities (e.g., NORDUnet, German, and Polish initiatives). Collectively, we look forward to reducing the isolation of our colleagues in the FSU through the facilities of the Internet. ================================================ n-2-1-013.30 Israel by Hank Nussbacher Israel has upgraded its internal ILAN backbone to 128kb/sec leased lines. In addition, it has also upgraded its link to PSI in the USA to 128kb/sec (satellite connection), giving Israel a total of 192kb international capacity (the other 64kb is to Europe). ================================================ n-2-1-013.95 Iranian Mullahs on the Internet? According to a recent press release from Iran's news agency, INRNA, three thousand Iranian mullahs have been trained to use computers for research in Islamic sciences. The director of a centre introducing computers to theological schools in Iran's main centre of Shi'ite moslem learning of Qom said it planned to link up with international computer-based information networks. The mainstream of Shi'ite moslem clergy has generally resisted technological innovations over centuries, but Iran's Islamic leaders say this should change. ================================================ n-2-1-014.10.2 Some Thoughts on Computer Networking in sub-Saharan Africa by Paul Nash* Introduction Africa might need peace and stability, but it also needs e-mail. While the telecommunications infrastructure is poor, there is an ever-growing need to communicate. More and more non-government organisations (NGOs) are sending workers, often with existing computer skills, into the field to assist with health-care, developing infrastructures, etc. These workers need to communicate with each other and with the various aid agencies funding them. Telephone lines are often few and far between, and line quality shocking. Fax transmissions seldom come through, and time-zone differences can make voice communication difficult. However, electronic mail makes good use of limited bandwidth, and the various transfer protocols allow for error detection and/or correction and restarting transmission. Store-and-forward facilities allow a number of shorter, more reliable local hops, leaving the intercontinental links to those with more reliable telecommunications. Due to the poor quality of phone lines, only one brand of modem is of any use in most of this continent: the Telebit Trailblazer and derivatives. While I hate to endorse any vendor's products, PEP beats the various CCITT modulation schemes hands down, providing up to 400cps over lines that will not support V.22bis. UUCP/UUPC The easiest way to transport Internet-format mail over erratic dial-up links is to use UUCP in its various flavours. Among its advantages are its ubiquity, with versions available for most Unix platforms, MS-DOS, Apple Mac and many other platforms. The protocol, however, shows its age. Even with Telebit's UUCP spoofing, throughput is not as high as it could be, and failed transfers are not restartable. Nonetheless, a simple laptop computer running MS-DOS can use a combination of Pegasus Mail and UUPC to provide a simple, cost-effective mail system. Fidonet Fidonet, the hobbyist network, uses an addressing scheme far removed from Internet standards. However, it normally uses Z-modem as the transport, and so gains in throughput what it loses in interoperability. For this reason, Fidonet technology networks have made great inroads into the African e-mail arena, and have become the de facto standard of most NGOs. Gateways between Internet mail and Fidonet can be complex, especially when they are to be kept as transparent as possible. I use RFMAIL under Unix to gate between the two networks, with a multiplicity of kluges to convert Fidonet addresses to Internet (and vice versa). My personal opinion is that the gains at the transport layer are seriously outweighed by the incompatibilities at application level. Current connections There are an ever-growing number of links into Africa. While I cannot describe all of them (and probably am not aware of half of them), I can provide an idea of the current status. Probably the most heavily used is the Greennet/Worknet network, which starts with Greennet in the UK, and uses Fidonet protocols to reach down through Africa, ending at Worknet in South Africa. From Worknet, traffic passes through the RFMAIL gateway to reach the South African universities' network (Uninet) and the rest of the world. Uninet also has a Fidonet gateway, which provides connections to the University of Zambia, as well as dial-up UUCP links to the Universities of Mozambique (Maputo), Namibia and Zimbabwe. There are also UUCP dial-up links between my Unix machine and Mozambique (Nampula), Zambia and Zimbabwe. These use Distnet, a PC package developed by SKAN Communications in Canada, and U-Access, a Macintosh package. There are further links on the cards to Angola (once the peace dies down) and places further afield. Experiences The experience so far has been extremely heartening. E-mail has proved a great success, and the demand seems to be growing. However, the telephone lines prove a greater and greater problem. Recently, the Zimbabwean PTT changed its international telephone exchange, and the University found that only two out of over two thousand attempted calls to South Africa succeeded. The link from northern Zambia to South Africa succeeds about one in four, with throughputs of around 100 cps (from modems that should provide 18000). In addition, although South Africa has reasonably good communications, the PTT here is opposed to "third-party" traffic, and tries to prevent such cooperative ventures. Although this may seem less than perfect, the current situation is a great advance on anything that went before. There is, however, room for a vast number of improvements. Future developments The field is ripe for alternative transport mechanisms, which can avoid the current shortage of telephone cables. Two that spring to mind are satellite and radio, and both should have a place in network plans. Satellite links can be most economically provided with micro-satellites, which act as flying mailboxes, while whizzing by in low orbit. The earth-station requirements for these machines can be met for as little as $5000. Due to their design, mail must be sent on a store-and-forward basis. There is at least one geostationary satellite over Africa, with bandwidth to spare. One of the ex-USSR military satellites now has bandwidth for sale, for those who can afford the $750000 C-band earth station and space-segment costs. Lastly, humble HF or VHF radios should not be ignored. These are cheap to provide, and while providing a comparatively low bandwidth, can provide real-time delivery. Conclusion To be competitive, UUCP needs a more efficient transport mechanism, such as Z-modem or Ian Lance Taylor's UUCP-i, to be used to deliver e-mail to non-Unix computers, and close the gap with Fidonet technology. A simple user interface, such as Pegasus Mail, allows neophyte users to operate a system with minimal training. Finally, consideration should be given to communications media that are not dependent on copper wires, as these are in short supply. *Communications Software Developer, Free Range Computer Systems ================================================ n-2-1-015.28 630 msec to McMurdo February is an active time in Antarctica. Lots of needed work is done before the bad weather sets in. Recently, an Internaut received an e-mail message from a technician at the satellite tracking station in McMurdo, Antarctica, who noted that they recently gained access to the INTERNET. Whereupon he proceeded to traceroute to mcmvax.mcmurdo.gov traceroute to mcmvax.mcmurdo.gov (157.132.103.50), 30 hops max, 40 byte packets 1 dellrouter.dell.com (143.166.224.2) 30 ms 10 ms 10 ms 2 UT6-S6.sesqui.net (128.241.4.161) 20 ms 10 ms 0 ms 3 UT1-E2.sesqui.net (128.241.0.241) 20 ms 10 ms 20 ms 4 RICE2-S5.sesqui.net (128.241.3.129) 50 ms 20 ms 20 ms 5 ENSS.sesqui.net (128.241.0.94) 30 ms 20 ms 10 ms 6 t3-1.Houston-cnss65.t3.ans.net (140.222.65.2) 20 ms 10 ms 20 ms 7 t3-3.Houston-cnss64.t3.ans.net (140.222.64.4) 20 ms 20 ms 20 ms 8 t3-2.Los-Angeles-cnss16.t3.ans.net (140.222.16.3) 40 ms 50 ms 50 ms 9 t3-2.San-Francisco-cnss8.t3.ans.net (140.222.8.3) 50 ms 70 ms 60 ms 10 t3-0.San-Francisco-cnss9.t3.ans.net (140.222.9.1) 70 ms 80 ms 70 ms 11 t3-0.enss144.t3.ans.net (140.222.144.1) 130 ms 60 ms 100 ms 12 ARC1.NSN.NASA.GOV (192.52.195.2) 80 ms 110 ms 140 ms 13 * ARC5.NSN.NASA.GOV (192.100.12.5) 60 ms 70 ms 14 128.161.115.2 (128.161.115.2) 660 ms 640 ms 620 ms 15 157.132.100.15 (157.132.100.15) 640 ms 640 ms 630 ms 16 mcmvax.mcmurdo.gov (157.132.103.50) 650 ms 630 ms 650 ms Indeed McMurdo is connected! [Please conserve their bandwidth and don't do this yourself.] ================================================ n-2-1-020.17 Medical Informatics The State University of Campinas, Campinas, SP, Brazil is noted for Internet activism. It hosts the worldwide biodiversity efforts. It also hosts a global initiative for medical informatics. Dr. Renato M.E. Sabbatini operates the Center for Biomedical Informatics from the University and publishes the Electronic Newsletter on Medical Informatics As part of the Center's activities, it maintains an on-line catalog of public-domain medical software for IBM-PC compatibles. It offers two kinds of software: Free-copy public-domain software, which was acquired from other sources. It can be download by anonymous FTP from the host (143.106.1.5) Shareware medical software. This software is protected by copyright and thus has a small price tag attached to it. It was developed by the staff of the Center for Biomedical Informatics of the State University of Campinas. The only way to acquire it is by sending a check in US dollars of UNESCO coupons to the address below. The diskettes are sent by airmail. In addition to the software, medical personal around the world can subscribe to Dr. Sabbatini's newsletter by sending a one-line message containing: SUBSCRIBE NIBNEWS Your Name to LISTSERV@CCSUN.UNICAMP.BR The shareware includes advanced tools in multiple languages, including: A shell program for generic building of rule-based expert systems in Medicine. The program is used in Medical Informatics and Artificial Intelligence courses at under- graduate and graduate levels in the institution. An educational program for teaching medical image processing techniques. A powerful, but simple to use, bibliographical reference manager. References are entered by means of a common word processor, and fields are tagged as in MEDLARS. A simple problem-oriented medical record package. An educational program for teaching digital ECG processing techniques. A neural network emulator for research and educational purposes. An educational software package for Windows 3.1, developed using Visual BASIC several common physiological simulations Dr. Sabatini can be reached at: SABBATINI@CCVAX.UNICAMP.BR or SABBATINI@BRUC.BIT NET ================================================ n-2-1-020.20 Experimenting with Library of Congress Classification in GopherSpace by Billy Barron* In my last column, I discussed the CICNet Electronic Journal Project. Recently, we have been conducting an experiment into the use of Library of Congress Classification for categorizing electronic journals in our Gopher server. Why Library of Congress Classification (LCC)? We picked LCC because the majority of the CIC libraries use it. Over the past few months, I have been involved in numerous discussions about the taxonomy of subject-oriented Gophers. The conclusion I have drawn is that there are numerous ways to classify materials and all have their relative merits. With some people though, the choice seems to be a religious issue. One of the problems with LCC is that some related topics are widely separated from each other. This is natural in any classification scheme that places an item in one place, which is necessary for physical holdings. To work around this problem, we have decided to add Gopher links to tie together related topics. For example, in LCC, Computer Science is in QA75 and QA76 and Computer Networks are in TK5105.5 to TK5105.9. In the CICNet Gopher, I plan on putting an entry under QA75-76 pointing at TK5105.5-9 to make finding the information easier. Another important aspect of this project is the building of a catalog and a searchable index to all the journals. I have yet to work on this aspect of the project, but it should prove to be interesting. I should stress that the use of LCC is just as a demonstration and that I am not a cataloguer. Therefore, I realize that some items are classified incorrectly. If you look through the collection (gopher.cic.net) and can correctly recatalog some of the items, drop me a note (billy@cic.net) and I will correct it. Anyway, it should be an interesting experiment. *VAX/Unix Systems Manager, University of North Texas ================================================ n-2-1-020.23 Network News from the American Mathematical Society--AMS by William B. Woolf * and David L. Rodgers** e-MATH--Services for Mathematicians The American Mathematical Society (with funding from the National Science Foundation--NSF) provides a number of services to the mathematical community through its e-MATH. Access is currently available to anyone with VT100 terminal emulation on Internet, by typing "telnet e-math.ams.org" or "telnet 130.44.1.100"; both the login name and the password are currently "e-math" (lower case). Among the services provided are: o the Combined Membership List of the American Mathematical Society, the Mathematical Association of America, and the Society for Industrial and Applied Mathematics; o a library of public-domain software related to TeX, AMS-TeX, and AMS-Fonts; o an employment information service, including all job listings which appear in the AMS publication "Employment Information in the Mathematical Sciences", plus the capability that job-hunters may post a brief curriculum vita which will be distributed to the employers with job listings; o access via Gopher to a number of other services on the Internet, including preprint and directory services; o listings of future AMS meetings; o access to an author-lookup service based on the files of Mathematical Reviews; o an electronic version of the Mathematics Classification Scheme; o an electronic version of The Bulletin of the American Mathematical Society; o a WAIS database containing items from the AMS Catalogue. AMS developing e-journal platform Work is under way at the AMS (with NSF support anticipated) on the development of a software platform to support the creation and distribution of electronic journals with advanced features including support of cooperative authoring, utilizing version-control and annotation capabilities. SGML and Mathematics--A DTD for mathematics fragments A subcommittee of the American Association of Publishers (AAP) convened by W. B. Woolf (AMS) is at work on the development of a DTD (Document Type Definition) for mathematics within SGML (Standard Generalized Markup Language). It is hoped to find a DTD which synthesizes the best features of an existing draft standard from AAP and an existing International Standards Organization standard (ISO 9573 TR 1988-1). A series of meetings including representatives of Euromath, Elsevier, ISO, CERN, Springer-Verlag, AMS, etc., has culminated in the creation of a small technical working committee which hopes to have a revised draft available for broad distribution late this winter. People interested in the technical issues involved can subscribe to the discussion list SGML-MATH by sending the message "subscribe sgml-math your_name" to "listserv@e-math.ams.org". *Associate Executive Director, AMS **Manager, System Development, Mathematical Reviews ================================================ n-2-1-030.50.2 Gigabit Networks by Craig Partridge, This column is about the different types of media that we can use to transmit data at gigabits per second. The first type of media is optical fiber. Briefly, an optical fiber is a strand of glass about the size of a human hair. The glass is specially treated so that at certain frequencies of light, light travels through the glass with very little loss. As a result it is possible (at least in theory) to send signals hundreds or even thousands of kilometers through a single piece fiber. (I should quickly note that these wonderful properties are a characteristic of single mode fiber. There's another type of fiber, called multimode fiber, which is easier to install and terminate, but has less good propogation properties). The bandwidth of a fiber is limited by the number of frequencies of light that we can transmit through it. Currently technology allows us to send about 75 TeraHertz (THz) through a single fiber, and we can signal between 1 and 1.4 bits per Hz, so a single fiber has a theoretical bandwidth of over 100 terabits per second. How are we going to actually send data at those bit rates? One way is to try to make the fiber into a wire with a big bandwidth. That's the approach used by Synchronous Optical Network (SONET), a telephony protocol for signalling over fiber. (SONET is part of a larger suite of CCITT standards known as the Synchronous Digital Hierarchy or SDH). In brief, SONET is a scaleable protocol that sends frames of data over the fiber. A SONET frame is made up of a number of repeated 90*9 chunks of bytes. To scale SONET up to higher speeds, one simply sends more chunks in each frame, while keeping the frame rate constant. At the lowest SONET speed of 51.8 Mb/s (called Optical Carrier or OC-1), the SONET frame is one 90*9 chunk. At OC-3 (155 Mb/s), the SONET frame is three 90*9 chunks; at OC-12 (622 Mb/s) the frame is twelve chunks, and so on. One problem with treating the fiber as a big wire is that developing fiber optic components that can signal at very high data rates is difficult to do. Currently, it is not possible to find components that signal at rates faster than OC-48 (2.4 Gb/s), which is a very small use of a 100 terabit per second fiber (OC-2,000,000 anyone?). One way to solve the challenge of combining slow optical components and large fiber bandwidths is to divide up the fiber frequencies into different channels, and have different components transmit at different wavelengths. This practice is known as Wave Division Multiplexing, or WDM. It is a sort of radio-in-fiber approach, where if your host and my host wish to communicate, we have to tune our hosts to the same channel. The key challenges in WDM are first, trying to pack the frequencies as close together as possible without having signals at one frequency interfere with signals at another frequency (a practice known as dense WDM), and second figuring out how to make sure that your host and my host can correctly (and quickly) find the right channel to tune to. IBM has demonstrated a WDM network called RAINBOW that supports several 300 Mb/s channels, and its designers hope to expand the RAINBOW design to support up to 1,000 channels, each capable of 1 gigabit per second. But what if you don't have access to fiber between two points? Can you still get gigabit performance? Indeed you can. If there's line of sight between the two points, then you can use a radio link. The Luckynet testbed at AT&T Bell Labs has an experimental 2.4 Gb/s (OC-48) radio link that covers a 23 mile distance between two labs. If you don't have line of sight, then perhaps satellites will come to the rescue. NASA is launching the Advanced Communications Technology Satellite (ACTS) in June. ACTS can connect multiple sites in the continental US at 622 Mb/s (OC-12) rates. Experiments with ACTS over the next two years will test how to effectively use satellite communications links at very high data rates. To learn more about fiber optic networking, consult Paul Green's new book, Fiber Optic Networks (Prentice Hall, ISBN 013-319492-2). Biswanath Mukherjee wrote a wonderful two part survey on the design of WDM networks in the May and July '92 issues of IEEE Network Magazine. The Luckynet radio link is described in a paper at the 1991 Globecom Conference. ================================================ n-2-1-030.65 U.S. President Holds Town Meeting on the Internet During a recent visit of President Clinton and V.P. Gore at Silicon Graphics the proceedings - incuding the President's announcement of his new National Information Infrastructure initiative - were distributed live over the MBone Internet as an audio-video multicast. Clinton's speech was followed by a Presidential "town meeting". ================================================ n-2-1-030.66 Internet Talk Radio by Carl Malamud On March 31, I'm launching a new service on the Internet called Internet Talk Radio. Internet Talk Radio is a "radio" metaphor: professionally produced radio programs that show up on the net as audio files. You can multicast them, or you can simply FTP the files and play. All I'm doing is producing the information and you are free to distribute at will using the protocol of your choice and to change the encoding format of the data to suit your computing platform. Distribution starts from UUNET and fans out to regional networks in an attempt to try and avoid excessive duplicate transfers. If you're a local net, you should contact your service provider. If you're a service provider, send mail to info@radio.com and I'll send you back instructions. If you're in Europe, mcsun at EUnet will be the initial spool point. If you're in Japan, WIDE and IIJ will do distribution. If you're a Alternet customer, you'll simply anonymous FTP from UUNET. We are not using the MBONE, although the networks that constitute the MBONE is certainly welcome to use that distribution medium if they feel that it is appropriate. The first show is "Geek of the Week" (;-), an interview show with members of the community. The program will be around a half-hour (e.g., 15 Mbytes in standard PCM, 8000 sample, 8 bit, mu-law encoding). The program is sponsored by Sun Microsystems and O'Reilly & Associates. Before the ugly spectre of AUP violations flames up .... we use a National Public Radio-style ack scheme consisting of just a couple of sentences. Indications from at least two of the large government networks are that we are compliant with their Appropriate Use Policies. --ed. A more lengthy description of Internet Talk Radio can be found in the February issue of ConneXions magazine. ================================================ n-2-1-030.80 International Character Sets in the Generic Network Services by Borka Jerman-Blazic The support of the International Character Sets in the forthcoming IT systems used through the network services is probably one of the most crucial requirements expressed by European users on many international forums. In that context, RARE decided to contribute to the area by acting as a focus for user-requirements in the emerging Character Set Technology and ensuring ways and methods these requirements to be met in the relevant standard documents and IT applications. The activities are carried out within the RARE Working Group on Character Sets Technology and the COSINE project. The group produced several papers about the problems and issues related to the use of International Character Sets in the network services. The group decided the problems of Character Sets should be worked out within two working levels (i.e., definition of the Character Sets Repertoires for RARE users and its application in the network services). Two services were identified as most important and relevant to the problem of Character Sets issues: MHS/X.400 and Directory/X.500. A document/report was produced which offers solutions to the use of International Character Sets in the Message Handling Systems. The document is in phase of being published as a RTR (RARE Technical Report) and is referenced in many international events in the field as an IETF Internet-Draft. Another document was also produced which defines the mapping between the Multipurpose Internet Mail Exchange protocol to MHS/X.400 providing solutions for gatewaying Internet mail containing message bodies coded with International Coded Character Sets standards to X.400 systems. The problem of the Directory and the support of International Character Sets is quite complex, and for that reason it was decided within the group that this subject is to be addressed in several steps. The major identified items which are planned to be worked on within the Task Force groups of the RARE Working Group on Character Sets set up recently within the new RARE Technical Program are the following: Definition of the Repertoire (mainly based on the CCITT T.61 recommendation and subrepertoires of ISO 10 646 to be derived from ISO 646 in collaboration with CEN TC 304). Definition of transliteration and conversion rules/tools for the Character Set Code Tables used by the majority of users in RARE community (i.e., ISO 646 compatible, ISO 8859 family, 8-bit EBCDIC family, 8-bit IBM PC code pages, 8-bit Macintosh, Hewlett-Packard ROMANS8, etc.). Promotion of DUAs and DSA of PARADISE, with the defined tables and developed conversion tools. The group is planning in the future to work on the Internationalization problems within networking by introducing pilots or Task Forces for special issues, in particular the following items are planned to be worked out: Character Set Technology, i.e., identification, coding methods, transformation/ conversion, correspondence between characters and glyphs, visual presentations at the device/user interfaces, and registrations via the Internet Assigned Numbers Authority (IANA). Promotion/piloting with the new Document Processing and Interchange Technology with support and use of different character sets repertoires and fonts, i.e., SGML, ODA, ODE, EDI/EDIFACT, and DSSSL. Evaluation of standards under development and providing input regarding user requirements. ================================================ n-2-1-040.31.1 Resource Discovery and Privacy by Mike Schwartz* As resource discovery and information services proliferate on the Internet, a number of privacy issues arise. The most obvious examples concern directories of people. What rights should people have to control what information about themselves is visible, who can access this information, and how the information is updated? Privacy problems arise in other types of directories as well. For example, from time to time users of the archie FTP directory service discover information that was intended only for limited distribution. Typically, this happens when a user looking for a particular program or document happens across other information while browsing the FTP site where the needed information was located. This happened, for example, with an early draft of a document being created by a networking organization last year. Another privacy problem arises in conjunction with the provision of directory service: access logs (which are routinely collected by many Internet services) can be used to determine peoples' interests, relationships, and other sensitive information. For example, logs of an individual's use of archie or Gopher could indicate interests in particular personal discussions. While it has always been possible to collect such information from older, non-directory oriented services (such as USENET news servers), the current generation of directory services can collect information about a much larger, more widely distributed collection of network users. It is typical for such services to receive requests from thousands of users around the world every day. In some cases, an explicit directory is not even needed to violate privacy. For example, it is possible to determine shared interest relationships between thousands of people by briefly monitoring and analyzing electronic mail "To:/From:" logs from a handful of sites. It is difficult to protect against this sort of invasion, because the needed data are not easily masked. Privacy Enhanced Mail does not protect against this type of analysis, because the actual content of mail messages (which is what PEM protects) is not needed for the analysis. Protecting against this type of traffic analysis would require generating spurious traffic, to hide patterns in the real traffic. Doing so could be costly. In some cases, users become more concerned about privacy when directory information becomes more widely or easily accessible. This is the case with Campus Wide Information Systems that get connected to the Internet, allowing people all over the world to locate information that previously was available only to people within a university. These problems also underlie some tension that has surrounded the Netfind user directory service. Originally, "finger" information was accessible only within local campus internets. As campuses connect to the Internet, this information becomes accessible to anyone with Internet access. Netfind makes it easy to harness this widely distributed information, and use it as a directory service. Because of this, some people have suggested that Netfind poses a privacy invasion, even though it provides no information that was not already publically available. What can be done about these privacy problems? A common approach is to enact security barriers, to protect sensitive information. Such barriers help, but often the goals of privacy and security are not well matched. The priority in most security systems is to protect a site's computer systems. Personal privacy is only protected to the extent that it overlaps with this goal. Yet, in most of the above examples, privacy could be threatened without violating a security perimeter. Moreover, in some cases privacy and security are at odds with one another. Every system administrator can tell a story of a time when, in an effort to track down a security problem, they faced a decision about whether to look into a personal mailbox for clues. Security mechanisms represent policies that have been formalized to the point of explicit controls over what users can do. Often, less formalized policies exist, in the form of proclaimed constraints on acceptable activities. For example, a university computing support group might have written policies describing the conditions under which a system administrator is allowed to inspect a personal mailbox. Still less formal policies exist concerning what type of behavior is acceptable when accessing machines across the Internet. For example, it is considered acceptable to login and retrieve files from a known remote anonymous FTP file system, but it is not considered acceptable to try to discover anonymous FTP servers by systematically attempting to login to a list of machines. In this case, the policy is really no more than conventional wisdom that has developed over years of collective Internet usage experiences. Sometimes this type of information is written down, in the form of new user guides. Often, however, it exists simply as folklore. A difficulty with defining privacy policies is that no governing body has the authority to legislate and enforce global policies. While a number of groups have stepped forward to suggest policies, there are no global privacy standards. Indeed, different countries currently have very different privacy laws and mechanisms. Even within a country, one can see a range of different directory privacy policy choices. For example, in the United States one extreme is occupied by mailing list brokers, which compile lists of everyone they can locate (assisted by the U.S. Post Office), and offer no assured way for people to modify or restrict access to the information listed about them. Telephone companies have a more moderate policy, listing everyone by default, but allowing individuals to restrict the content or presence of their listings. Still more moderate are the policies put forward by the North American Directory Forum (NADF). NADF advocates several principles, including the right to be informed when a person's directory entry is created, the right not to be listed, the right to correct inaccurate information, and the right to remove specific information. Notice that none of the above policies requires that users give prior consent to be listed - which is often what people upset about privacy intrusions really want. Even the NADF policy allows users to enforce their privacy preferences only after the information has been visible for some initial period of time. Given today's technology, even a short period of time may be enough to spread private information to many derived databases. There are at least three reasons why privacy policies tend to be weaker than individuals would like. The first is ubiquity of coverage. A user directory is only valuable if it contains listings for many people. Requiring prior approval makes this goal nearly unattainable. The second problem is more vexing: invading privacy can be very profitable (as in the case of direct mail marketing lists). Third, governments often want to maintain more detailed information about people than their citizens would like, in the name of national security. There are no easy solutions to privacy problems. However, I have three suggestions. First, we need to educate users about the many ways that networked information can invade their privacy. In a sense, networks represent an "electronic society", participation in which brings with it some loss of privacy, just as being a member of any society does. Second, we need to increase the amount of research and development oriented towards ensuring privacy. Technical conferences on the subject often focus almost entirely on system security, with few papers about privacy. Third, we need to form policies oriented towards current technology. The world has changed substantially since the U.S. Privacy Act of 1974 was introduced, and that legislation does not effectively address many of the privacy problems that face us today (nor is it effectively enforced). One possibility would be for the U.S. National Research Council to commission a study parallel to last year's security study, focused on privacy problems raised by networked information. Or perhaps the Internet Society could commission such a study, being an international organization. There are many other privacy problems introduced by electronic data processing, beyond issues raised by Internet directory and information services. The interested reader can contact one of the offices of the Computer Professionals for Social Responsibility (the main office being in Palo Alto, California), or proceedings of the recent conferences on Computers, Freedom & Privacy. (-ed. References and citations in this article are available upon request from Prof. Schwartz.) *Professor, Computer Science Department University of Colorado. ================================================ n-2-1-040.33.1 Towards an Internet Security Architecture: Part III by Stephen Kent* kent@bbn.com This is the third installment in a multi-part series addressing architectural security issues in the Internet. The columns in this series explore various aspects of an Internet security architecture as the community begins to explore these issues. Some of the material in this column reflects observations contained in the chapter on Architectural Security in the "Internet System Handbook" published by Addison Wesley in October 1992. This installment continues the exploration of security services, and security mechanisms used to provide these services, using the terminology introduced in ISO 7498-2, the OSI security architecture. As noted previously, security services are abstract characterizations of security functionality, independent of the underlying mechanisms used to implement the functionality. Thus security service definitions provide a helpful vocabulary for end users, and network or application service providers, to express security requirements. These definitions also serve as a means for protocol and application designers to express the security functions provided by protocols and applications or required of an underlying communication system. The second security service to be explored is that of integrity. Data afforded the integrity service is protected from undetected, unauthorized modification. ISO 7498-2 further characterizes integrity as connection-oriented or connectionless, depending on the communication protocol context for which the service is provided. The distinctions between these two forms of integrity have real implications both in terms of the security features provided to the client of the service and in terms of the mechanisms used to implement the service. In some ways the integrity security services are analogous to the integrity services one expects of a reliable a communication protocol. However, a security-oriented integrity service provides protection in the face of malicious attacks against data whereas a reliability-oriented integrity service provides protection against benign errors, accidents, acts of God, etc. Connectionless integrity is applied to individual packets or messages. This service strives to provide to the recipient of a packet (message) with confidence that the received data has not been modified enroute. Note that a simple error detection code, e.g., a CRC, is not, by itself, sufficient to afford this security service. The reason is that a malicious attacker is presumed to be capable of modifying a packet and computing a new CRC with a "correct" value. This is an example of why the security mechanisms developed for benign error situations are generally not sufficient to provide protection against malicious attacks. Instead, integrity mechanisms for use in hostile environments must be able to protect against attackers who are presumed to be capable of performing any algorithmic transformation, e.g., checksum calculation, that the sender or receiver can perform. The only "leverage" afforded the legitimate participants in a communication is the knowledge of some secret quantities which may be shared between the sender and receiver, or which may be linked to publically known quantities (as in public-key cryptography). Numerous algorithms been developed for checking the integrity of individual packets exchanged between a pair of communicating parties, many of which are termed "manipulation detection codes." Examples are codified in the ANSI X9.9 specification and the SNMP security enhancements (e.g., RFC 1352). For multi-cast environments, more stringent techniques are required, to counter the possibility that one of the legitimate recipient might attempt to modify a message and pass it on to other recipients. The Privacy Enhanced Mail (PEM) specifications (RFCs 1421-24) describe a means of providing integrity in such environments. Connection-oriented integrity extends the connectionless integrity guarantee to include detection of "any modification, insertion, deletion, or replay of any data within a [packet] sequence," according to ISO-7498-2. This defines a fairly strict integrity service well suited to a range of applications, e.g., virtual terminal and file transfer. Such applications require that absolutely all packets be delivered in sequence and without error. However, this service characterization does not represent the extreme on a spectrum of services. Applications such as packet speech or packet video will function acceptably well even if some packets never arrive, or are damaged. However, these applicatins are sensitive to variability in delays. Thus integrity can be viewed as a multi-dimensional security service. This motivates the more general concept of "stream integrity," which attempts to encompass a wide range of integrity services which may be appropriate for varying applications. Stream integrity also seems a preferable service description since not some of the applications which require more than simple connectionless integrity do not make use of connection-oriented protocols. Mechanisms for providing stream integrity are quite varied. No single stream integrity mechanism is ideal, or even appropriate, for all applications, because of different application integrity requirements. Stream integrity techniques usually include the connectionless integrity techniques alluded to above, then add features such as non-repeating sequence numbers, timestamps, challenge-response exchanges, etc. Finally, in isolation, integrity is not a very valuable security service. When an individual packet or message or a stream of data arrives, the recipient usually wants to know not only that the data has not been modified in any way, but also who transmitted the data. This latter service is referred to as authenticity and is the topic of the next column. *Chief Scientist BBN Communications ================================================ n-2-1-040.33 Privacy and Security Research Group Workshop on Network and Distributed System Security by Jeffrey I. Schiller*, On February 11th and 12th, 1993 the Privacy and Security Research Group held the first PSRG Workshop on Network and Distributed System Security. Co-sponsored by the Internet Society and the Lawrence Livermore Laboratory, the workshop was held in San Diego, California and attracted approximately 150 networking professionals from around the globe. The workshop was divided into two types of presentations, paper sessions and panel sessions. Papers ranging from policy making to technical dissertation were presented. Topic areas covered Privacy for Large Networks, Electronic Documents, Privacy Enhanced Mail (PEM), and Practical Distributed System Security. The panel sessions were designed to present the attendees with an opportunity to discuss timely and in some cases controversial security topics. Panels included the question of in which layer(s) of the OSI reference model security should be implemented in. Smart cards and their applications were discussed in another panel. One panel discussed exportable encryption algorithms (and what their use implied). The closing panel asked the question: "Should security be legislated?" (i.e., should workstations have required security features in the same way that automobiles and airliners must meet safety standards). Although each panel had several presenters, group participation of all workshop attendees resulted from the extended question and answer periods. The workshop Banquet featured Scott Charney, Chief Computer Crime Bureau of the U.S. Justice Department, going over the challenges facing law enforcement when it comes to apprehending and prosecuting computer criminals. Among the issues covered he included a discussion of the motivation behind the CERT's (Computer Emergency Response Team) recent advice to warn network users of keystroke monitoring, a topic that has spawned significant debate within the network community. Scott explained that quirks in U.S. law rather then a desire to "spy on the community" is behind the advice. The session on Electronic Documents had papers presented on secure electronic meeting management and secure electronic document handling. Both papers presented ways to use networking, and the Internet, to handle business/administrative tasks in a secure fashion. The PEM session speakers presented papers covering the security issues of implementating PEM in a multi-user UNIX environment and VAX/VMS PEM implementation and user experiences. An implementation of a hardware certificate signing unit was presented, as well as a paper outlining some of the subtle security issues (and solutions) for implementing PEM with symmetric key cryptography. The session on distributed systems discussed practical applications of security technology, to the problems of distributed systems and shared file systems. All in all, the workshop was well attended and extremely productive. The papers were excellent and the panel discussions enlightening. Current plans call for this workshop to become an annual Symposium. It is a Symposium worth investigating. Be on the lookout for the next Call for Papers! *MIT Network Manager, Massachusetts Institute of Technology ================================================ n-2-1-040.35 How Reliable are PTT International Links? by Hank Nussbacher Nowadays, countries order high speed data communications links for hundreds of thousands of dollars. For example, Israel spends $207,000 per year on a 128kb satellite line to the USA and another $140,000 per year on a 64/kb fiber link to Europe. But just how reliable are those expensive circuits to abroad? The CCITT (International Telegraph and Telephone Consultative Committee) has stated that intercontinental circuits should not experience more than 3 hours per month down time (99.6% up time) and that continental links should not experience more than 1 hour down time (99.9% up time). Unfortunately, a recent INTUG (International Telecommunications User Group) report that monitors various PTTs throughout the world, shows that on average 64kb circuits are available only 99.03% (7 hours per month downtime). How has Bezek, the Israeli monopoly PTT, been doing compared to its European cousins? The table below uses two international circuits for the basis of its analysis: a satellite 64kb (128kb as of 26 January 1993) circuit from Weizmann Institute in Rehovot to New York City (USA) and a second 64kb circuit that goes via the EMOS, a fiber optic undersea link, from Tel Aviv University to Geneva. This table shows that Bezek is more or less on par with around 99.1% uptime for international circuits, but which is clearly below the CCITT recommendation of 99.6% uptime. IDB, the USA carrier for the other half of the USA-Israel circuit, states in their contract that reliability will not be less than 99.5%. Clearly, they do not reach that objective. Israel-USA | Israel-Europe (satellite) | (EMOS - fiber optic cable) | | Month Downtime # of % | Downtime # of % in min. failures avail | in min. failures avail ------ -------- -------- ----- | -------- -------- ----- Jul 92 462 20 99.0 | 532 24 98.9 Aug 92 657 68 98.5 | 300 15 99.3 Sep 92 424 47 99.1 | 276 14 99.4 Oct 92 100 25 99.8 | 230 89 99.5 Nov 92 524 26 98.8 | 297 31 99.3 Dec 92 28 17 99.9 | 134 37 99.6 Jan 93 568 32 98.7 | 1114 88 97.5 What are the major causes of a digital telecommunications circuit to go out of service? Synchronization and multiplexing signals on high bandwidth international links is more complicated than on slower speed analog links. Synchronizing at the international level is particularly tricky because different carriers (PTTs) often use different signalling schemes. These synchronization problems usually result in a link disruption of under 2 minutes, long enough to force users off a remote application. For example, over the period of October 3rd & 4th, the circuit from Israel to Geneva had 64 line failures, each on average 18 seconds in length. Fortunately, we use equipment that requires only 10 seconds to recover full user connectivity in the event of a line failure. But someone who would be using certain vendor equipment that would require a reboot after every line failure, may experience 20-30 minute outages for every 18 second line synchronization problem. Due to the fact that different PTTs use different signalling standards, it is close to impossible for PTTs to monitor digital circuits and instead wait for customers to call them to report circuit outages. PTTs usually show higher availability statistics than customers, since they only start their "down clock" when a customer reports the problem, which is often 15-30 minutes after the line has gone down. Interestingly enough, the higher the data speed, the more reliable the digital data circuit (from the INTUG report): Speed Uptime ------ ------ 768kb 99.99% 384kb 99.94 1544kb (T1) 99.94 1024kb 99.9 128kb 99.88 1984kb 99.88 512kb 99.87 2048kb (E1) 99.56 64kb 99.03 256kb 97 Looking at the above table, users would be recommended to stay away from 256kb circuits and to upgrade as fast as possible from 64kb to 128kb circuits. Almost all higher speed lines (other than 64kb and 256kb) meet the CCITT recommendations. I would be interested in forming a group in the Internet to discuss these matters and to exchange statistics, and I am prepared to coordinate the matter as well as attempt to prepare coordinated reports on link uptime and reliability of lines based on country. Those interested in discussing this, should send me e-mail at the address stated above and if we get enough of a turnout, we might be able to improve the reliability of our leased lines. ================================================ n-1-4-040.51 User Services by Joyce K. Reynolds* On 5 January 1993, the National Science Foundation's (NSF) Network Information Services Awards were announced. Enclosed is an abbreviated announcement of this new and important award for the Internet community. In cooperation with the Internet community, the National Science Foundation developed and released, in the spring of 1992, Project Solicitation for one or more Network Information Services Managers to provide and/or coordinate (i) Registration Services, (ii) Directory and Database Services, and (iii) Information Services for the NSFNET. As a result of this solicitation, three separate organizations were competitively selected. Together, these three awards constitute the NIS Manager Project, named the INTERNIC. Network Solutions will provide registration services, AT&T will provide directory and database services, and General Atomics will provide information services. The three awardees have developed a detailed concept and plan to provide this seamless interface called the "INTERNIC" and have agreed to the structuring of their three separate awards as one collaborative project. Network Solutions will provide registration services as the IP registrar, issue IP numbers worldwide using delegated registries under the guidance of the Internet Assigned Numbers Authority and also register domain names, and track points of contact. Applications for assignment will be accepted via email or facsimile. The information from these assignments will be provided to the directory and database services provider to be made available to the entire Internet community. As a part of the Domain registration efforts Network Solutions will periodically release the top level zone files to be used by all root Domain Name servers. AT&T will develop and maintain a Directory of Directories, including lists of FTP (File Transfer Protocol) sites, lists of various types of servers available on the Internet, lists of white and yellow page directories, library catalogs and data archives. AT&T will also provide white and yellow pages type Directory Services. Access to these services will initially be provided through several currently popular in-use interface methods while migrating to the use of X.500 technology, the current standard specification for distributed information storage and retrieval. The database services which AT&T will provide include the establishment of Database Services to extend and supplement the resources of the NSFNET, such as extend and supplement the resources of the NSFNET, such as databases of contributed materials of common interest to the user community. AT&T will also offer database design, management, and maintenance to institutions and groups for inclusion in the Internet. General Atomics will provide Information Services acting as the NIC of first and last resort and the NIC of NICs. The INTERNIC information services will include a full-service Reference Desk, a database of comprehensive networking materials called the Info Source, training classes and documentation, and coordination services among all appropriate groups in the community. In keeping with the innovative spirit of the Internet, several new approaches to distributing services will be implemented. Among these innovations is NICLink, a user-friendly hypermedia interface offering access to the Info Source and all the information it contains. NICLink will be distributed on both standard computer diskettes and CD-ROM. Another is the concept of the Info Scout, an individual who will scout out new resources and innovative uses of the network for inclusion in the Info Source. National Science Foundation Contact Network Solutions Contact Don Mitchell Mary Bloch (202) 357-9717 (703) 742-4740 dmitchel@nsf.gov maryb@netsol.com AT&T Contact General Atomics Contact Shelly London Susan Calcari (908) 221-4355 (619) 455-3900 london@attmail.com calcaris@cerf.net *Member of the Technical Staff, Information Sciences Instititue, University of Southern California ================================================ n-2-1-040.68 All You Didn't Want to Know About BITNET and Were Afraid To Ask by Eric Thomas BITNET's store-and-forward technology may not deliver the best round trip figures for electronic mail, but there is one area in which it outperforms current Internet technology by a large factor: bulk mail delivery. This article will show that, contrary to popular belief, this difference is not made irrelevant by upgrading your access line to T1. In a previous article ("What's the Response Time", by Frode Greisen and Greg Lloyd), we saw that the average round-trip time for a typical 50-lines mail message over the EARN backbone is on the order of 300-400 seconds, or about 3 minutes for the one-way trip. Even though these figures include downtime and some of the countries on the EARN backbone do not have satisfactory connectivity, this is still pretty unimpressive. Even between well-connected countries, the average transmission time would be on the order of 30-90 seconds during peak hours, assuming the lines are up. Between the same machines, SMTP delivers the same message in 5 seconds, maybe 10 if the machines are loaded. So why aren't the BITNET folks working on dismantling this outdated technology and migrating everything to IP? Well, we computer people can compare figures and argue about the conditions of our benchmarks to our heart's content, but the reality is that users don't care whether a message is delivered in 5 seconds or 2 minutes. It took them 5 minutes to type it, usually with two fingers, and as long as it gets there within some 10-15 minutes they will be satisfied. That is, genuine end-users aren't going to judge a technology solely on whether it takes 5 seconds or 2 minutes to deliver a piece of mail; this will be just a minor factor among many others in their overall evaluation of the service. Now, while Joe Bitnetter is willing to wait 10-15 minutes for his message to be delivered, he of course expects the message to be delivered within that 10-15 minutes period regardless of whether it was sent to a single person or to a mailing list. After all, the network handles millions of messages a day, so a couple thousand additional messages should make no difference. And this is, indeed, the way the post office works: a thousand letters mailed first class take the same time to arrive as a single one. And here it is where the Internet fails to deliver. The round trip time observed on the IETF list (from a T1-connected site) ranges from 30 minutes to several hours. The message gets to CNRI in less than one minute, but takes about an hour to come back. BITNET, on the other hand, uses a bulk delivery algorithm known as "DISTRIBUTE", which uses knowledge of the topology to optimize the distribution in a store-and-forward fashion. There are about 150 DISTRIBUTE servers spanning 28 countries, and the distribution algorithm ensures that a single copy of the message is sent over any given "link" (originally these were actual, physical leased lines, nowadays they are mostly virtual TCP circuits between designated machines). For instance, a LISTSERV somewhere in California could forward a copy of the message to a server on the East Coast, which would fan it out to a number of other East Coast servers and send a copy to Stockholm, where the message would be delivered to recipients in Sweden and then passed on to Poland and Finland - and so on. A study has shown that, on the average, this reduces the load by a factor of 5 to 25, depending on the topology and usage patterns. Note that 35-40% of BITNET mailing lists are highly-specialized interest groups with around 100 recipients or less: the savings are of course higher with larger lists. The bandwidth saved through this algorithm is clearly a gold mine to countries with a weak economy, such as eastern countries, which simply haven't got the money to purchase fast lines (leased lines usually have to be paid for in hard currency). With two 64k connections, Poland is by far the best connected eastern country, and that was accomplished (partly) through subsidies. If a particular software technology can reduce the bandwidth used up by e-mail by a factor of 10, you will have a hard time convincing countries connected at 9.6k or even 64k that this is a minor side-effect of an obsolete technology they should strive to migrate away from. Unless, of course, you get them a free ride on a T1 in exchange. In addition to this purely financial benefit, we come to another significant aspect of this distribution mechanism: it evens out the cost of delivering messages to mailing lists of more than a couple hundred subscribers. It doesn't make much of a difference for the network backbone whether there are 500 or 10,000 recipients, because the message will have to be delivered to (almost) all the participating DISTRIBUTE servers in both cases. The difference is that there will be a lot more deliveries on local area networks and faster/cheaper regional lines, which usually aren't backlogged. This is why BITNET passes the "post office" test in regions without chronic bandwidth problems (and considerably eases the load on regions which do have continuous backlogs). To quantify that point, we conducted an experiment with a moderately large but pretty active list (about 1,200 recipients and over 100 daily postings), hosted in the US. We selected two "guinea pig" accounts, one in the US (but in a different State) and one in Poland. Both accounts were subscribed three times to the mailing list, under different incarnations: once with the machine's BITNET address, once with an Internet address for which a "site gateway" had been registered in the BITNET gateways database, and once with an Internet address for which no particular gateway was defined (in fact we used source-routing in the second case to simulate the behaviour we wanted, since the sites in question had chosen not to register any "site gateway"). We were careful to add the addresses in the middle of the list file, so that they wouldn't get unfair (or "too fair") treatment. We then recorded the arrival time of the various "triplets" on a 24 hour cycle, to compare the relative efficiency of the 3 delivery methods: 1. BITNET/NJE all the way. 2. BITNET/NJE to the "site gateway", then SMTP to the final mailbox. 3. SMTP all the way. In all but one case, the first method was the fastest. The round trip time (between 'Date:' time stamp and last 'Received:' field) was on the order of 10 minutes for the US recipient and 15 for the Polish one, give or take a couple minutes (the clocks of the various machines were not synchronized and we had no time to check the offset of the individual clocks of each and every poster on that day). On the average, the copy sent to the "type 2" address arrived 1 minute 30 seconds later for the US recipient and 4 minutes 40 seconds later for the Polish recipient. However, the copy with the plain "type 3" Internet address (SMTP all the way) arrived more than an hour after the other two, for both addressees. Now, this might be excusable for the Polish address, since the IP route to this host is usually saturated, but what about the T1-connected US university? What would they tell Joe (ex-)Bitnetter if they were to leave BITNET and he suddenly had to wait an extra hour for his copy of all postings to medium-to-large mailing lists? On active lists, getting everything 1 hour after everyone else may effectively cut you from the argument: by the time your reply arrives, it is already 2 hours older than what the BITNET folks have posted and you are probably just repeating what other people said before - and giving the unfortunate impression that you don't read what has been said before replying. Of course, one way to address this problem is to buy faster lines and bigger machines, thus making it possible for SMTP servers to handle 500 concurrent TCP connections and 5,000 outstanding name server queries, and then run name servers which can take 20,000 simultaneous queries without timing out a single one. The technology to handle such workload exists, even though cost and other factors have caused it not to be widely deployed. But why not use smaller machines that do a good job because they only need to deliver a dozen copies of a message, rather than machines which are so big that they can handle the full 1200, every third minute during peak hours, for each of the hundreds of large mailing lists we now have? As the network and its community grows, there is no limit to the further resources a "non-DISTRIBUTE" delivery system will require. LISTSERV is delivering 2-6 million messages a day without hassle - without anything special to do when creating a large list, and without having to worry about finding a suitable newsgroup to move the discussion to once the list grows large. The distribution is efficient, not only in terms of bandwidth and computer resources (about one second of CPU time per thousand recipients on the smallest machines you can buy from IBM, rated at just 3 S/370 MIPS), but above all in terms of manpower. The system takes care of itself and you don't have to press people to set up local redistributions to avoid saturating your machine or getting complaints from your local network or system administrator. While this distribution mechanism is implemented over BITNET, it can, like most other LISTSERV services, be used from the Internet without problem. In fact, you can take advantage of DISTRIBUTE to optimize the turnaround time of your sendmail exploders, simply by entrusting the delivery to the DISTRIBUTE backbone, as explained in informational RFC 1429 (Listserv Distribute Protocol). Furthermore, a number of projects are underway to improve the efficiency of this distribution mechanism for Internet recipients (for which it is difficult to obtain accurate topological information). LISTSERV is being modified to support "network aware" mode in Internet-only environments; that is, within a few months it will be possible to install fully functional LISTSERV servers on VM systems without BITNET connectivity, and these servers will be able to participate to the DISTRIBUTE backbone. Another project concerns itself with the collection of topological data for the Internet, without which there can be no efficient distribution. The EARN Association, in particular, is going to review the possibility of collecting, updating and maintaining such information in a standard form as a value-added service to its membership. ================================================ n-2-1-040.70 USENET by Rick Adams [graphics separately provided} ================================================ n-2-1-040.90 News from RARE by Josefien Bersee START-UP CONTRACT OPERATIONAL UNIT SIGNED BY RARE The setting up of the Operational Unit for the provision of networking services to the European research community is progressing. RARE - acting on behalf of the OU - has signed the contract with the CEC for the start-up of the OU, which will be located in Cambridge in the UK. RARE TECHNICAL COMMITTEE REAPPOINTED The RARE Technical Committee, the expert body guiding all RARE's technical activities since April last year has been reappointed for a two years' term, ending 1 May 1995. In addition to the existing members, Milan Sterba (Prague School of Economics) was designated to represent RIPE. Currently there is one vacancy, as Eike Jessen had to resign. The current RTC is chaired by Tomaz Kalin, RARE Secretary-General; Tim Dixon, RARE Project Development Officer acts as Secretary. Other members are: Brian Gilmore (Edinburgh University, United Kingdom), Erik Huizer (SURFnet, Netherlands), Jean-Paul Le Guigner (CICB, France), Bernhard Plattner (ETH Zuerich, Switzerland) and Sven Tafvelin (NORDUnet, Chalmers Technical Institute, Sweden). Howard Davies (RARE Vice-President, University of Exeter, UK) liaises between the RTC and the RARE Executive Committee and Christian Huitema (INRIA, France) liaises between RARE and the Internet Architecture Board. PUBLICATION OF RARE/EARN JOURNAL During its last meeting in Luxembourg in February, the RARE Council of Administration approved the budget for the production of a paper version of a new journal jointly published by RARE and EARN. The new publication will appear as quarterly supplement to the Elsevier North Holland Journal, "Computer Networks and ISDN Systems". The first issue is planned to be published in Spring 1993. PRELIMINARY PROGRAMME 4TH JENC AVAILABLE The preliminary programme and registration form for the 4th Joint European Networking Conference is now available from the RARE Secretariat, on paper as well as electronically. The conference will be held in Trondheim, Norway, at the campus of the Norwegian Institute of Technology. NEW PROJECT DEVELOPMENT OFFICER AT RARE SECRETARIAT Jeroen Houttuin has started work as PDO at the RARE Secretariat in Amsterdam on January 1st. He studied telematics at the University of Twente and has been involved in R&D networking projects at GMD-FOKUS, ETH Zuerich, and SWITCH. He participated in the COSINE MHS Project Team. Jeroen will be involved in RARE Working Group activities, upper-layer related issues in particular. ================================================ n-2-1-050.01 NEW ADMINISTRATION BACKS INFORMATION SUPERHIGHWAYS by Mike Roberts At a well publicized meeting in Silicon Valley in late February, President Bill Clinton and Vice President Al Gore announced a technology initiative for their new administration with potentially up to US $20 billion in funding over the next several years. Contained within the plan is a major initiative for "Information Superhighways" that is intended to build on Vice President Gore's previous support for a National Research and Education Network (NREN) and expand it to include investment in the information infrastructure of the entire U.S. Specific investments are targeted for networking support in education at all levels, in manufacturing technology, health care, and lifelong learning. Although the announcement covered many areas of interest to the networking community, details are still sparse. The Administration will not release its detailed budget for FY94, which begins 1 October 1993, until the first week of April. Many important positions in the federal executive necessary to carry out the initiative are still vacant. Controversy has arisen over whether the Clinton Administration intends that the federal government have an operational role in a national high performance computer network. Many American telecommunications firms are anxious to avoid any expansion of federal activity in markets they consider already too regulated. At the same time, public interest advocates are concerned that excessive reliance on market forces will exclude many citizens from network access, thus impacting their ability to acquire necessary work skills in an increasingly information based economy. In the Congress, both House and Senate members have introduced legislation dealing with aspects of networking, ranging from telecommunications policy to the assignment of network responsibilities to several federal agencies. Both the U.S. computer industry and the university community have called for the appointment of a high level official to manage the large scale networking effort being undertaken by the new Administration. Thus far, this step is being resisted by forces within federal agencies and their Congressional sponsors whose prerogatives might be adversely affected as a result of putting a single individual in charge. ================================================ n-2-1-075.01 by Deborah Estrin Internetworking: Research and Experience is a quarterly, refereed journal published by Wiley (D. Comer, R. Droms, D. Estrin, and L. Svobodova are the editors.) In the second 1993 issue of the journal there are two papers of interest to the ISOC community. The papers represent the diversity of research in the field, from modeling to protocol design and experimentation. "Model and Analysis of a Virtual Circuit with Cross Traffic", by B. Jain (IIT), A. Agrawala, and Dheeraj Sanghi (Univ. Maryland), models the flow of packets between a source and destination host, while taking into account processing and competing cross-traffic in the network elements (switches). The model's predictions of inter-packet departure time as a function of inter-packet send time are analyzed and compared to measurements taken between host pairs on the Internet. "Inter-Domain Routing Protocol" by Y. Rekhter (IBM), describes and analyzes a protocol for inter-domain routing based on a Path Vector algorithm. Path vector protocols are distributed in the sense of distance vector protocols, however they avoid loops by including the full domain-level path in each routing entry, in addition to the usual (destination, distance) tuple found in most distance vector protocols. IDRP has been designed to accommodate networks of virtually unlimited size; for example, aggregation of routing information is widely deployed, and at the same time, flexible aggregation is widely supported. The protocol was developed originally for OSI routing but can be applied to IP routing as well. ================================================ n-2-1-075.04 ConneXions by Ole J Jacobsen* The March 1993 edition of ConneXions is a special issue devoted to INTEROP 93 Spring (March 8 - 12 in Washington, DC). This edition starts with an article on the INTEROPnet, INTEROP's unique show network to which all exhibitors are required to connect and demonstrate interoperability. This is followed by and article on APPI, the alternative to IBM's APPN. OSF's Distributed Computing Environment (DCE) is described next, as well as the soon-to-debut Internet Talk Radio. Finally you will find an article entitled "Multiprotocol Internets: User's Best Hope for Open Systems." The article describes several important trends in the networking marketplace. We are pleased to offer all Internet Society members a 20% discount on ConneXions subscription. Indicate your membership number on the order form. For a complimentary sample issue, list of back issues and subscription information, please send e-mail with your POSTAL address to connexions@interop.com. Future INTEROP dates and locations: ----------------------------------- INTEROP 93 Fall: San Francisco, CA, August 23 - 27, 1993 INTEROP 93 Europe: Paris, France, October 25 - 29, 1993 NETWORLD + INTEROP 94 Las Vegas, NV, May 2 - 6, 1994 NETWORLD + INTEROP 94 Atlanta, GA, September 12 - 16, 1994 *Editor and Publisher ConneXions--The Interoperability Report Interop Company ================================================ n-2-1-075.05 Computer Networks and ISDN Systems by Philip H. Enslow, Jr.*, Computer Networks and ISDN Systems, Published by North Holland, Amsterdam. As reported previously, Computer Networks and ISDN Systems is now published twelve times a year. We are well along into the 1993 volume with issue number 7 having just been received by subscribers. Volume 25, Number 7, February, 1993 is a special issue on, "Tools for FDTs" with Juan Quemada as the Guest Editor. Plans are underway to significantly increase the coverage of research networking in Europe. More on that in the next issue of ISOC News. *Editor-in-Chief, "Computer Networks and ISDN Systems" ================================================ n-2-1-075.06 Matrix News by John S. Quarterman* jsq@tic.com ]*indicates italics*] The December 1992 *Matrix News* included a technical overview of ``MIME: Multimedia Across the Internet,'' by Mark Thacker, an essay on ``Paying for Internet Goods and Services...,'' by Peter Deutsch, and an extensive list of ``Recent Internet Books,'' by John Quarterman, with a table comparing the books by length, type, price, intended audience, and type (e.g., user guide, travelog, or textbook). You'll probably find your favorite book in the list. If not, let us know. We're always publishing more reviews and adding to the list. The January 1993 issue contained much of the paper trail Gordon Cook compiled in his investigation of ``NSFnet Privatization,'' by use of the Freedom of Information Act (FOIA). This is extended excerpts from the actual documents, with very little editorializing. In the same issue, Eric Theise reviewed the book ``Zen and the Art of the Internet,'' by Brendan Kehoe, and Smoot Carl-Mitchell reviewed two items from InfoMagic: a CD-ROM with many documents and software, and a set of floppy disks. Both have RFCs; the disks have them in Hypertext. We also ran the announcement by NFS of the ``Network Information Services Awards'' and the announcement by EFF of ``Major Changes at the Electronic Frontier Foundation.'' For February 1993 we have a special *Matrix News* issue with one article, ``Maps of Networks by Country.'' This article consists almost entirely of bar graph maps, with a bar over each country or region, showing the number of hosts on a network in that area. There are maps for each of FidoNet, UUCP, BITNET, and the Internet. Each network is shown in each of the world, Europe, and the United States, for twelve maps total. The March 1993 issue of *Matrix News* will have some material on the most networked countries, the most northerly and southerly networked places, and a bit of a graphical surprise, in color. *Matrix Information and Directory Services, Inc. (MIDS) tel:+1-512-451-7602 ================================================ n-2-1-100.08 IFIP News by Howard L. Funk IFIP mourns the death of its key founder, Mr. Isacc L. Auerbach, who died of myelofibrosis on 24 December 1992. He was instrumental in the creation of IFP and served as its first president from 1960 to 1965. We owe him an immeasurable debt. FOCUS, the United Representative to IFIP has elected new officers. Dr. Robert M. Aiken - Chair; Dr. Mario R. Barbacci - Secretary/Treasurer: Dr. Stuart H. Zweben, Dr. J. Tom Cain, and Prof. Gordon B. Davis - Directors. Through the good offices of the Internet Society information about IFIP and FOCUS now lives on GOPHER. The ftp repository is software.watson.ibm.com in the /pub/ifip directory. The listserve repository is in LISTSERV@CEARN (request index txt to see what is available). Last September the IFIP Technical Assembly approved the formation of a Working Group on Human-Computer Interaction and People with Special Needs. For additional information contact the Chair of this group, Porf. J. Gonzales-Abascal (julio@si.ehu.es). The Council of European Professional Informatics Societies (CEPIS) met in Vienna in March. IFIP decided to hold its Council meeting in Vienna during the same week to give the leadership of both organizations the opportunity to meet to discuss future relationships. The following meetings may be of special interest to members of the Internet Society. Third International Symposium of Integrated Network Management, 18-23 April 93, San Francisco, USA. Ninth International Conference on Computer Security, 12-14 May 93, Toronto, Canada. Thirteenth International Symposium on Protocol Specification, Testing, and Verification, 25-28 May 93, Leige, Belgium. First Working Conference on Development and Implementation of Computer-Based Information and Communications Networks, 16-18 June 93, Vienna, Austria. For additional information about any of these events contact the IFIP Secretariat: ifip@cgeueg51.bitnet. ================================================ n-2-1-900.07 Internet Research Task Force (IRTF) by Jon Postel* The research groups of the IRTF are oriented towards longer term Internet research issues. The research groups tend to be small (that is, about 10 to 20 people), and to have consistent membership. The results of the research groups tend to be experiments in new protocol ideas, new applications, and new approaches. Any results of research groups that are appropriate for wide spread use in the Internet are submitted to the normal IETF standards process. Many of the participants in the research groups are also active in the computer science research community and there is a healthy interaction between the work in the research groups and work in other venues (such as ACM SIGCOMM). The IRTF is composed of the following research groups: Research Group Name Chair ------------------- ---------------- Autonomous Networks Deborah Estrin (USC) End-to-End Services Bob Braden (ISI) Resource Discovery Mike Schwartz (U Colorado) Privacy and Security Steve Kent (BBN) Libraries Cliff Lynch (UCOP) The IRTF Steering Group (IRSG) is composed of the research group chairs (above) and the following at large members: Dave Clark (MIT) Dave Mills (UDEL) Bruce Schatz (U Arizona) *Associate Director for Networking HPCC Division, Information Sciences Institute, University of Southern California ================================================ n-2-1-900.08 Note from the RFC-Editor by Jon Postel* Request for Comments documents (RFCs) are now available via FTP from at least 20 on-line repositories around the world, and at least 6 of these provide automated retrieval via email. A complete listing of these sources and other "help" information about accessing RFCs can be obtained via an email request to the RFC-INFO service: To: RFC-INFO@ISI.EDU Subject: Accessing RFCs Help: ways_to_get_rfcs In the last five months, 61 RFCs have been published. Oct Nov Dec Jan Feb 10 10 1 27 13 The most recent summary of the status of various standards and their corresponding RFCs is "IAB Official Protocol Standards" which is STD-1 and RFC-1360. *Associate Director for Networking HPCC Division, Information Sciences Institute, University of Southern California ================================================ n-1-2-900.09 The Internet Assigned Numbers Authority (IANA) by JonPostel* The IANA is the central coordinator for the assignment of unique parameter values. Requests for parameter assignments should be sent to . The most recent summary of these assigned parameter values is "Assigned Numbers" which is STD-2 and RFC-1340. *Associate Director for Networking HPCC Division, Information Sciences Institute, University of Southern California ================================================ n-2-1-900.12 Standards Actions by Steve Coya* The Internet Engineering Steering Group (IESG) approved or recommended the following twenty-two (22) actions between 1 January 1993 and 28 February 1993: o Network Time Protocol as a Draft Standard. o A String Representation of Distinguished Names as a Proposed Standard. o Using the OSI Directory to Achieve User Friendly Naming as an Experimental Protocol. o Privacy Enhancement for Internet Electronic Mail: Part I: Message Encryption and Authentication Procedures as a Proposed Standard. o Privacy Enhancement for Internet Electronic Mail: Part II: Certificate-Based Key Management as a Proposed Standard. o Privacy Enhancement for Internet Electronic Mail: Part III: Algorithms, Modes, and Identifiers as a Proposed Standard. o Privacy Enhancement for Internet Electronic Mail: Part IV: Key Certification and Related Services as a Proposed Standard. o Definitions of Managed Objects for the DS1 Interface Type now has a status of Historic. o Definitions of Managed Objects for the DS3 Interface Type now has a status of Historic. o Definitions of Managed Objects for the DS1 and E1 Interface Types as a Proposed Standard. o Definitions of Managed Objects for the DS3/E3 Interface Type as a Proposed Standard. o FYI on a Network Management Tool Catalog: Tools for Monitoring and Debugging TCP/IP Internets and Interconnected Devices as an Informational Document o SMTP Service Extensions as a Proposed Standard. o SMTP Service Extension for Message Size Declaration as a Proposed Standard. o Transition of Internet Mail from Just-Send-8 to 8Bit-SMTP/MIME as an Informational Document o SMTP Service Extension for 8bit-MIMEtransport as a Proposed Standard. o Mapping between X.400(1984/1988) and Mail-11 (DECnet mail) as an Experimental Protocol. o Lightweight Directory Access Protocol as a Proposed Standard. o The String Representation of Standard Attribute Syntaxes as a Proposed Standard. o Internet Users' Glossary as an Informational Document o Directed ARP as an Experimental Protocol. o DUA Metrics as an Informational Document Forty (40) Requests for Comments (RFC) were published between 1 January and 28 February 1993: RFC St Title ------- -- ------------------------------------- RFC1384 I Naming Guidelines for Directory Pilots RFC1387 I RIP Version 2 Protocol Analysis RFC1388 PS RIP Version 2 Carrying Additional Information RFC1389 PS RIP Version 2 MIB Extension RFC1390 S Transmission of IP and ARP over FDDI Networks RFC1391 I The Tao of IETF: A Guide for New Attendees of the Internet Engineering Task Force RFC1392 I Internet Users' Glossary RFC1393 E Traceroute Using an IP Option RFC1394 I Relationship of Telex Answerback Codes to Internet Domains RFC1395 I BOOTP Vendor Information Extensions RFC1396 I The Process for Organization of Internet Standards Working Group (POISED) RFC1397 PS Default Route Advertisement In BGP2 And BGP3 Versions Of The Border Gateway Protocol RFC1398 DS Definitions of Managed Objects for the Ethernet-like Interface Types RFC1401 I Correspondence between the IAB and DISA on the use of DNS throughout the Internet RFC1402 I There's Gold in them thar Networks! Searching for Treasure in all the Wrong Places RFC1403 PS BGP OSPF Interaction RFC1404 I A Model for Common Operational Statistics RFC1405 E Mapping between X.400(1984/1988) and Mail-11 (DECnet mail) RFC1406 PS Definitions of Managed Objects for the DS1 and E1 Interface Types RFC1407 PS Definitions of Managed Objects for the DS3/E3 Interface Type RFC1408 PS Telnet Environment Option RFC1409 E Telnet Authentication Option RFC1411 E Telnet Authentication: Kerberos Version 4 RFC1412 E Telnet Authentication : SPX RFC1413 PS Identification Server RFC1414 PS Ident MIB RFC1415 PS FTP-FTAM Gateway Specification RFC1416 E Telnet Authentication Option RFC1417 I NADF Standing Documents: A Brief Overview RFC1421 PS Privacy Enhancement for Internet Electronic Mail: Part I: Message Encryption and Authentication Procedures RFC1422 PS Privacy Enhancement for Internet Electronic Mail: Part II: Certificate-Based Key Management RFC1423 PS Privacy Enhancement for Internet Electronic Mail: Part III: Algorithms, Modes, and Identifiers RFC1424 PS Privacy Enhancement for Internet Electronic Mail: Part IV: Key Certification and Related Services RFC1425 PS SMTP Service Extensions RFC1426 PS SMTP Service Extension for 8bit-MIMEtransport RFC1427 PS SMTP Service Extension for Message Size Declaration RFC1428 I Transition of Internet Mail from Just-Send-8 to 8Bit-SMTP/MIME RFC1429 I Listserv Distribute Protocol RFC1430 I A Strategic Plan for Deploying an Internet X.500 Directory Service RFC1431 I DUA Metrics Key to RFC Status: S Internet Standard PS Proposed Standard DS Draft Standard E Experimental I Informational * Executive Director, IETF Secretariat ================================================ n-2-1-fill1 Silicon Valley Realtors on the Internet A recent issue of the Palo Alto Weekly carried a full-page ad for a Realty Company in Saratoga California. Like most such advertisements, it has a photograph of each of the 35 realtors who work for the agency. Under each photograph is an Internet address of them. ================================================ n-2-1-fill2 4th JOINT EUROPEAN NETWORKING CONFERENCE The 4TH Joint European Networking Conference will take place in Trondheim, Norway, from May 10-13, 1993. Theme of the conference is: "EUROPEAN RESEARCH NETWORKING IN A GLOBAL CONTEXT" The event is organized by RARE in cooperation with: ACM SIGCOMM, EARN, EUnet/EurOpen, IFIP TC6, Internet Architecture Board, Internet Society, NORDUnet, and UNINETT. The conference will be held at the main campus of the Norwegian Institute of Technology The programme includes a wide variety of technical, operational, and developmental topics. This year for the first time tutorials will be organized, and held at the end of the Conference. The registration fee includes the full set of papers as they will be presented during the conference, and all festivities. For more information: RARE Secretariat, Josefien Bersee (bersee@rare.nl) ================================================ n-2-1-fill3 _The Internet Society's On-Line Information Services_ by John W. Stewart III The Internet Society has taken steps to make it easier for people to access information of interest to ISOC members. Information is available via Gopher, anonymous FTP, and WAIS servers, all at ietf.cnri.reston.va.us. (An introduction to Gopher can be found via anonymous FTP on boombox.micro.umn.edu under: /pub/gopher/gopher_protocol/protocol.txt.) The Gopher is currently configured to provide information about the Internet Society, Internet Engineering Task Force, International Federation for Information Processing and, as a special bonus, US White House press releases. For all three of these organizations, the Gopher provides text files, gateways to anonymous FTP areas, and gateways for keyword searching of documents. The Internet Society News is accessible on-line through these services. The anonymous FTP files of particular interest are in sub- directories: iesg, ietf, ietf-mail-archive, internet-drafts and isoc. WAIS information databases of interest include INFO, isoc, rfc and internet-drafts. The Gopher server has been registered with the University of Minnesota, and can be accessed by choosing the "Internet Society" entry underneath the "North America / USA / General" listing. ================================================ n-2-1-fill4 Dynamic WAIS Prototype Announcement by Mike Schwartz Dynamic WAIS is a system that extends the WAIS (Wide Area Information Service) paradigm to support information from remote search systems (as opposed to the usual collection of static documents). We have a prototype that supports gateways from WAIS to Archie and to Netfind, using the Z39.50 information retrieval protocol to seamlessly integrate the information spaces. Unlike menu-level gateways (such as the current interim telnet interface from the Internet Gopher system to Netfind), Dynamic WAIS uses the Z39.50 information retrieval protocol to seamlessly integrate the information spaces. This approach allows users to access a great deal of diverse information without learning multiple user interfaces. The key idea behind the Dynamic WAIS gateways is the conceptual work of constructing mappings between the WAIS search-and-retrieve operations, and the underlying Archie and Netfind operations. In the case of Netfind, for example, when the Dynamic WAIS user requests a search using the "dynamic-netfind.src" WAIS source, the search is translated to a lookup in the Netfind seed database, to determine potential domains to search. The Netfind domain selection request is then mapped into a WAIS database selection request. Once the user selects one of the domains to search, the WAIS retrieval phase is mapped into an actual domain search in Netfind. The actual implementation of Dynamic WAIS involves fairly straightforward extensions to the standard WAIS code. Whereas standard WAIS has a search and a retrieval phase, with dynamic WAIS the retrieval phase can trigger a remote search (which could cascade if the retrieval for that search caused another search, although the current prototype only goes one level deep). The problem that arises is that the standard WAIS client does not send the keywords to the server during the retrieval phase. The Dynamic WAIS client sends the keywords in both phases, and also understands how to display the results of Dynamic searches. If you would like to learn more about Dynamic WAIS, you can obtain the source to the Dynamic WAIS prototype via anonymous ftp from ftp.cs.colorado.edu in /pub/cs/distribs/dynamicwais. This directory also contains WAIS ".src" files for the Dynamic WAIS gateways to Archie and Netfind. A paper on Dynamic WAIS is in preparation. This WAIS server was created in January 1993 by Darren R. Hardy and Michael F. Schwartz as part of the Networked Resource Discovery Project. ================================================ n-2-1-fill5 TCP/IP "Bake-Off" HELD TO ENSURE INTEROPERABILITY - --First Test Meet Sees 13 Vendors Testing Protocol Suite Implementations-- Engineers from 14 companies gathered at North Andover, Massachusetts, in February for a multivendor Transmission Control Protocol/Internet Protocol (TCP/IP) "Bake-Off". The vendors attending the Bake-Off exercised their implementations of the TCP/IP protocol suite to test interoperability between products. Participating in the week of testing were: BSDI, Data General Corporation, Digital Equipment Corporation, Empirical Tools and Technologies, Inc., FTP Software, Inc., Hewlett-Packard Company, InterCon Systems Corporation, Lachman Technology, Inc., Mentat Inc., Microsoft Corporation, Novell, Inc., TGV, Inc., Tule Network Services, and Xyplex, Inc. While there have been multivendor networks at trade shows such as UniForum, the Federal Computer Conference and INTEROP for nearly 10 years now, as well as groups who meet regularly to work with SNMP, NFS, X and similar standards, there has been little cooperative testing of the underlying TCP/IP stacks and their associated application-level protocol suites. This appears in the fact that there are some implementations currently available which are unable to successfully negotiate even some of the more basic communications options, or keep network links open when hardware or software anomalies occur. "Networking customers do - and should be able to - expect a wide range of products to be able to interoperate", said one of the organizers of the event. "Participation in a testing event such as this demonstrates a vendor's commitment to quality, and at the same time provides another potent quality control function for networking consumers." He went on to say that "network vendors are presented with an unique paradox: in order to compete against each other, they have to work together - because any networking product that cannot communicate with another networking product rather rapidly ceases to have discernable practical utility." The Bake-Off effort began when an open invitation to participate was given at the November 1992 IETF (Internet Engineering Task Force) meeting in Washington D.C., the Bake-Off was founded on an example set by Jon Postel of USC's Information Sciences Institute, one of the original TCP/IP designers, almost 12 years ago. While the recent testing took place in a secure area, and actual test results will not be made public, some of the tests designed and used by the participants will be made openly available to members of the networking industry. Since this was the first such event in so long a time, it was expected, in part, to be a trial effort. All participants felt that the exercise was a success, and there are plans to hold another Bake-Off within the year. In fact, several vendors have reported that improvements developed at the Bake-Off would be included in future releases. ================================================ n-2-1-fill6 Internet and the Handicapped For many of us, the Internet has become an indispensible part of our professional life and career activities. For other Internauts, it is somewhat more special. The following note was recently posted by David Richman on the Shakespeare Electronic Conference list: A query to anyone on this list associated with *Shakespeare Bulletin* or *Shakespeare Quarterly*. Any chance of either of these periodicals becoming available electronically, perhaps through GOPHER or some other electronic service. Being blind, I and my speech synthesizer take delight in online material. I would be willing to pay up to twice the normal subscription rate for either of these journals in electronic form. Thanks. ================================================ n-2-1-fill7 The Secret Internet As the Internet continues to scale exponentially worldwide, it is being constantly featured in numerous publications. One of the more unusual, however, was the 26 December issue of THE ECONOMIST. An articled entitled "THE GOOD NETWORK GUIDE - Being one of us" ranked networking groups on several factors. The Internet made this noted list, but with rather curious attributes. The following are measured on a scale from 0 to 5, 0 = lowest, 5 = highest: P = Power S = Secrecy O = Organization B = Strength of Beliefs P = Peculiarity of rituals E = Exclusivity Conspiracy ("Network") P S O B P E --------------------------------------------- --- --- --- --- --- --- Old Etonians 3 1 1 2 3 4 Cambridge University Conservative Association 3 1 4 0 2 2 Skull and Bones 2 4 2 0 5 5 Inspection Generale de Finance 5 2 5 3 1 5 Doon School 4 1 2 3 1 4 law school of Tokyo U. 5 2 5 3 1 5 Rhodes scholarship 3 0 3 1 1 4 Mont Pelerin Society 2 3 3 5 2 3 Committee to Defend the Workers (KOR) 4 3 3 4 1 4 The Communist Party (XSU) 4 5 2 1 1 3 Broederbond 4 4 4 4 4 2 Muslim Brotherhood 2 3 4 5 3 1 Opus Dei 2 4 4 5 5 1 Freemasonry 3 4 4 3 5 2 Trilateral Commission 4 3 3 1 0 5 USENET and Internet 2 0 1 2 5 1 Order of Illuminati 5 5 5 5 5 5 ================================================ n-2-1-fill8 ISOC Nominations Committee Report Six Trustee positions are to be determined in the 1993 elections. Four of these positions are Trustee positions created in 1993, and two positions are the result of the standing down of Trustees Harms and Aiso. Selected Candidates --------------------- The following individuals have been selected by the Nominations Committee as candidates in the 1993 election: Peter Bakonyi h25bak@huella.bitnet Xavier Baquero xbaquero@ecnet.ec Scott Bradner sob@das.harvard.edu Nevil Brownlee nevil@ccu1.aukuni.ac.nz Brian Carpenter brian@dxcern.cern.ch Susan Estrada estradas@cerf.net Dave Farber farber@cis.upenn.edu Howard Funk funk@vnet.ibm.com Haruhisa Ishida ishida@u-tokyo.ac.jp Gary Malkin malkin@xylogics.com Jean Polly jpolly@nysernet.org Dave Sincoskie sincos@thumper.bellcore.com Petition Candidates --------------------- The following individuals have submitted membership petitions signed by a minimum of 50 ISOC individual members by the nominated due date: Jon Postel postel@isi.edu Yakov Rekhter yakov@watson.ibm.com Candidate Details ------------------- Submitted details of all listed candidates are held on the host aarnet.edu.au in the directory /pub/isoc/candidates Biographies of all candidates are also available in the Internet Society Gopher server (ietf.cnri.reston.va.us) and also in the ftp/isoc/elections anonymous FTP directory on the same machine. The Internet Society Nominations Committee: G. Huston (Chair) R. Blokzijl I. Fuchs T. Kalin C. Partridge H. Tokuda ================================================ n-2-1-fill9 Transition to the New INTERNIC Team Please note, after 31 MARCH, operations of the NSF Network Service Center (NNSC) will be discontinued. The services formerly provided by the NNSC will be transferred to a new Network Information Services Management team, collectively known as the INTERNIC (the Internet Network Information Center). As a result of a competitive solicitation, the National Science Foundation (NSF) has awarded new contracts for NSFNET/NREN services: * Network Solutions (NSI), has provided registration for the NSFNET since January 1992, and will continue to perform these services. * AT&T will provide expanded directory and database services. * General Atomics, which now operates CERFnet and the San Diego Supercomputer Center, will provide a Help Desk and general information services. The new INTERNIC phone number is 800-444-4345 and email is info@internic.net. The phone number is scheduled to be in service on 1 April 1993 with a voice menu: 1 - REGISTRATION SERVICES Direct dial: 703-742-4757 Network Solutions, Inc. (NSI) Email: hostmaster@nic.ddn.mil Herndon, VA 2 - DIRECTORY AND DATABASE SERVICES Direct dial: 908-668-6587 AT&T FAX: 908-668-3763 5000 Handley Road, Room 2F25 Email: admin@ds.internic.net South Plainfield, NJ 07080 3 - INFORMATION SERVICES Direct dial: 619-455-4600 General Atomics Email: info@internic.net San Diego, CA 4 - TO SPEAK WITH A RECEPTIONIST After April 1, the resource-guide and policies-procedures online files will be made available through AT&T/INTERNIC, as will the shadow copies of RFCs, internet-drafts, ietf, iesg, and isoc. The NNSC Info-Server will no longer be available, but the INTERNIC hosts will provide email servers. The nnsc.nsf.net will announce details as they become available. A longer version of this announcement is in /nsfnet/transition in the anonymous ftp directory on nnsc.nsf.net, or send email to info-server@nnsc.nsf.net with the these lines in the text field: request: nsfnet topic: transition ================================================ n-2-1-fill10 INET'93 - Towards a Global Community International Networking Conference Internet Society San Francisco, CA 17-20 August 1993 The International Networking Conference is the annual conference of the Internet Society, a new professional society serving the Internet community. Following the very successful INET'92, INET'93 will be held on 17-20 August 1993 in San Francisco. Focusing on worldwide issues of research and academic networking, the goal of INET'93 is to bring together individuals from university, industry and government who are involved with planning, developing, implementing, managing and funding national, regional and international research, academic, and commercial networks. The conference agenda will include plans and status reports on research and academic networks throughout the world. Major topics include: Network Technology: Advances in the Network Technology Base Network Engineering: Building the Global Infrastructure Application Technology: Enabling Technologies for Distributed Applications User Applications: Support for International Communities of Interest Regional Issues: Networking Around the Globe Policy Issues: Governance, Management, and Financing of International Networks The conference will be held immediately preceding Fall Interop '93, the leading trade show for Internet technologies. This will make possible attending both events as well as tutorials given as part of Interop '93. Workshop for Developing Countries A workshop designed to assist developing countries in their installation and use of networking technology and services is being organized and will take place during the week before the conference in the San Francisco Bay Area. Conference Chair: Eric Benhamou, President, 3Com To be added to the conference mailing list or for other requests, send mail, fax or E-mail to: USRA ATTN: INET'93 625 Ellis Street, Suite 205 Mountain View, CA 94043 USA tel: +1 415 390-0317 fax: +1 415 390-0318 Request@inet93.stanford.edu the INTERNIC (the Internet Network Information Center).