European International Academic Networking A 20 Year Perspective

Скачать презентацию European International Academic Networking A 20 Year Perspective

e1daa61f2deb186a34f383a132229098.ppt

Количество слайдов: 45

European International Academic Networking A 20 Year Perspective Peter T. Kirstein University College London June 7, 2004 TNC 2004 20 Years of European Academic Networking 1

Outline • • Background The Status in 1984 The Protocol Wars Network Consolidation European Connectivity Foreign Connectivity Advanced Services The Future June 7, 2004 TNC 2004 20 Years of European Academic Networking 2

Contents • • Background The Status in 1984 The Protocol Wars Network Consolidation European Connectivity Foreign Connectivity Advanced Services The Future June 7, 2004 TNC 2004 20 Years of European Academic Networking 3

Scope • Mainly Academic Service Networks – I. e. Networks for research, not network research – Little about Telecom networks unless for research • Concentrating on European Level – Mention US mainly for comparison – Not considering National activities • Covering mainly last two decades June 7, 2004 TNC 2004 20 Years of European Academic Networking 4

1984 - The US Scene • Agencies had mission-oriented networks – NASA DECnet-based SPAN, NSN (IP under consideration) – Do. E MFEnet (Fusion, proprieatary), HEPnet (DECnet), – DARPA Arpanet SATNET (/IP) – DISA Milnet (Arpanet protocols) – NSF CSNET (Arpanet Protocols, G/w to Arpanet • IBM – BITNET all IBM mainframes – many • USEnet, PHONEnet – both mail only June 7, 2004 TNC 2004 20 Years of European Academic Networking 6

1984 - European Scene • Few NRENs – Holland, Norway, Sweden, UK – COM in Norway and Sweden - COM • Many PTT X. 25 PDNs – too expensive • CSNET node in Karlsuhe U – IP etc • France many separate nets – INRIA, CNRS, University (still 6 -node Cigale) • UK JANET link to Arpanet via SATNET/IPSS – CB • SATNET service sites in Germany, Italy, Sweden and UK – only UK to NREN, others U only – IP etc • CERN large net and X. 25 switch for HEP - Prop – Many HEP labs linked and had inter-site traffic • European BINET (EARN) nearly ready – IBM Prop • Many used EUNET for mail - UUCP June 7, 2004 TNC 2004 20 Years of European Academic Networking 7

Early Research Protocols • Development of OSI Protocols vs Internet – TCP vs TP 4/CLNS or CONS, FTP, X. 400 vs SMTP, X 500, XXX vs TELNET, Cambridge Ring, Token vs ENET • Many tried to define and live with OSI family – Germany particularly keen – Many worked on X. 400 and X. 500 (84 & 88) • Looking at impact of PTT services – ISDN, X. 25, videotext, teletex, fax • Lot of work on gateways and adaptation • Some complete systems – particularly UK – File, mail, NRS, RJE, XXX, transport, LAN, CONS, X. 400, X. 500, Coloured Book June 7, 2004 TNC 2004 20 Years of European Academic Networking 8

The Protocol Wars • US nets mainly leased lines, European X. 25 • 7 layer OSI vs SNA vs DECNET vs Internet vs CB – Each layer generated arguments, including whether layer was at correct place or needed • Different Groups worked on Standards – – – – CCITT improved X. 25, X. 400, X. 500, XXX ISO File, Presentation, Alphabet, Transport IETF worked on Internet Protocols (from 1986) DEC on DECNET protocols IBM/BITNET/EARN on SNA and EARN JNT on Coloured Books Significant amount of cross-filing of standards June 7, 2004 TNC 2004 20 Years of European Academic Networking 10

Protocol Consolidation • During 84 -90, Protocols families mixed – Put BITNET, Internet, DECNET, CERNET, UUCP all above X. 25 – X. 400, X. 500 above TCP/IP – BITNET-2, DECNET above TCP/IP • Allowed Possibility of single network infrastructure with multiple families – EARN, OSI, Internet/X. 25 – BITNET, DECNET, X. 400, X. 500/TCP/IP June 7, 2004 TNC 2004 20 Years of European Academic Networking 11

Network Consolidation • • Background The Status in 1984 The Protocol Wars Network Consolidation European Connectivity Foreign Connectivity Advanced Services The Future June 7, 2004 TNC 2004 20 Years of European Academic Networking 12

US Network Consolidation • DARPA kept remit only for its researchers – Arpanet/SATNET de-commissioned in ’ 88 – DARTNET, Gigabit net, CAIRN etc replaced it • NSF started NSFNET in ’ 86 to connect five supercomputer centres at 64 Kbps – Quickly widened to whole research community – Moved to 1. 5 Mbps in 88, with 45 Mbps later – Set up regional nets in 88, kept foreign links Stopped NSFnet in 1995 • ESNET, NASA moved to IP nets in ‘ 86 -’ 90 June 7, 2004 TNC 2004 20 Years of European Academic Networking 13

US Network Linking • 88 -90 Set up Federal Interchanges under FNC and FRICC – Somewhat limited facilities in interchange – Allowed sharing of international connections – Kept separate physical infrastructures • Encouraged outsourcing of many services – NSF first regional then backbone – NASA eventually their NSN – NSF/DARPA the whole Gigabit net and then VBNS and finally Internet-2 June 7, 2004 TNC 2004 20 Years of European Academic Networking 14

Early European IBM Activity • In 84, EARN started as European BITNET – By 1/95, Denmark, France, Germany, Ireland, Israel, Italy, Spain and Switzerland – Usually one international node, national nodes had to be put in by institutions – Encouraged EARN/X. 25 to share lines • Some countries, particularly UK, did not want penetration; provided national gateways • Was very important for HEP, because US labs were on EARN and later EASINET • By 88, EASINET initiative went to IBM Supercomputers – was all TCP/IP • In 92 became part of EBONE, merging also with EUNET – June 7, 2004 TNC 2004 20 Years of European Academic Networking 15

Early Links of European NRENs • Most NRENs developed as X. 25 ones with OSI protocol suites • From around 84, started having bilateral links as needed • Links were expensive and slow, moreover European PTTs had international links between X. 25 packet data networks • EC had started Framework Programme in ’ 84 – needed links between organisations • EC commissioned PTTs to provide IXI in 87 – – 64 Kbps X. 25 backbone in 18 countries Free access to universities via NRENs Access charges for commercial organisations Lasted until 92, with changed prime June 7, 2004 TNC 2004 20 Years of European Academic Networking 16

Network Coordination - OSI • All mission oriented groups had coordination from beginning • RARE started in 1986 as a discussion group for NREN common problems – 1 st European Networkshop in 85, the run by RARE – Ran Cooperation for Open Systems Interconnected Networks in Europe for EC (COSINE) – COSINE funded IXI, X. 400, X. 500, Security Pilots – In 1990 had first joint workshop with EARN JENC June 7, 2004 TNC 2004 20 Years of European Academic Networking 17

Coordination of IP Activity • Internet had coordination from 83 – IAB and ICB from 83 – IETF etc from 86, first non-US in 90 in Amsterdan • Set up Reseaux IP (RIPE) in 89 for administration and technical IP coordination – Based in Amsterdam hence 1 st IETF there • By 93, European-wide networks to be procured – Set up Trans-European Research and Education Networking Association (TERENA) and Delivery of Advanced Network Technology in Europe (DANTE) – Both owned by NRENs June 7, 2004 TNC 2004 20 Years of European Academic Networking 18

Mission European – US Links • DARPA’s SATNET links had impact only for UK Internet – JANET CB service • ESNET connected to CERN and Garching via EARN, and DECNET to CERN X. 25 • NASA ran SPAN and later NSN to its researchers in several countries • NSF went from NSFNET to INRIA, Karlruhe and later Stockholm for NSFNET • Later all went to MAE-EAST ot MAE-WEST – Costs were borne by lead Federal Agency, but access proficed to all the federal nets – Most ran TCP/IP, bringing them into the European systems June 7, 2004 TNC 2004 20 Years of European Academic Networking 19

EC co-funded EUROPANET • Already mentioned IXI, by 90 inadequate • Both because of US links and LANs, many countries wanted some TCP/IP • By 92, PTTs were competing, and did not want to repeat IXI at higher B/w • Some NRENs wanted X. 25, some DECNET, some TCP/IP – In 92 European Multi-Protocol Backbone (EPMB) formed at 2 Mbps (X. 25, IP, DECNET) – later became Europanet – Included many countries – Links to be 2 Mbps June 7, 2004 TNC 2004 20 Years of European Academic Networking 20

EBONE Set Up • Some NRENs did not like EMPB, and wanted Pure IP • This group agreed to set up EBONE – Main nodes London, Montpellier, CERN, Amsterdam and Stockholm – Ireland, Spain, Nordunet, EARN and EUNET joined – US links provided by IBM from EASINET – Gateways to IXI and later EMPB in Amsterdam – Links 256 -512 Kbps June 7, 2004 TNC 2004 20 Years of European Academic Networking 21

Later Nets • By 1993, higher speed nets were needed, and PTTs liked ATM – RACE agreed to fund JAMES 2 Mbps ATM infrastructure – as a research pilot • By 1994, the protocol war was won, only IP survived. Future nets pure IP – TEN-34 started in 96 – 34 Mbps – QUANTUM (TEN-155) in 98 – 155 Mbps – GEANT in 00 2. 4 Gbpx – GEANT-2 in 03 10 Gbps June 7, 2004 TNC 2004 20 Years of European Academic Networking 22

EUROPANET June 7, 2004 TNC 2004 20 Years of European Academic Networking 23

TEN 34 – November 1997 June 7, 2004 TNC 2004 20 Years of European Academic Networking 24

QUANTUM – May 2001 June 7, 2004 TNC 2004 20 Years of European Academic Networking 25

GEANT BB – Dec 2002 June 7, 2004 TNC 2004 20 Years of European Academic Networking 26

NREN IPv 6 traffic volume June 7, 2004 TNC 2004 20 Years of European Academic Networking 27

Growth of IPv 6 Connections • GÈANT v 6 steady migration – Core Configuration – February 2003 – Pilot service – June 2003 – IPv 6 production service – October 2003 • 27 out of 31 NRNs connected – 24 out of 27 are native connections – 356 routes announced to them • US Abilene, ESNET, CA-NET, Koren peers June 7, 2004 TNC 2004 20 Years of European Academic Networking 28

Foreign Links • US always different – Up to mid-90 s, bilateral agreements – Now ordered by DANTE, amount varies, often direct from QUANTUM/GEANT • EARN had links to Israel, Turkey, Cyprus – Remained as networks became European • Long-time NREN links to, Canada, China, Japan, Korea, African France etc – Often Technical, sometimes political, e. g. G 7 TEIN June 7, 2004 TNC 2004 20 Years of European Academic Networking 30

EC Programmes • After break-up of Soviet Union, EC, NATO, Soros, UNDP had special programmes. The EC ones concentrate on connecting to European Nets – From 91, EC-PHARE to Bulgaria, Czech, Hungary, Poland, Roumania, Slovakia – From 94, EC-PHARE/INSIGHT Albania, Estonia, Latvia, Lithuania, Slovenia – TEN-34 already all PHARE, Ebone, Europanet, Japan and US – EUMEDIS includes Mediterranean countries – SEEREN treats the Balkan countries – ALICE goes to South America – TEIN-2 will include Southern Asia June 7, 2004 TNC 2004 20 Years of European Academic Networking 31

Other Programmes • NATO Science for Peace Network Panel deals with former East Europe, Russia and NIS – 94 -00 set up NRENs and connectivity to all above – 96 -02 VSAT system to HEP labs in Russia, Ukraine, Armenia, Georgia, Kazakhstan co-funded by DFN and DESY – 02 -05 SILK VSAT system to all NIS of Central Asia and S. Caucasus (with additional funding from EC, Cisco, DESY, Soros and others) • AID, IREX, Soros, UNDP, World Bank all have complementary development programs June 7, 2004 TNC 2004 20 Years of European Academic Networking 32

Advanced US Service Networks • When Arpanet stopped, DARPA researchers needed advanced services – DARTNET, CAIRN etc did research on advanced services like M/c and IPv 6, but were available also to DARPA researchers – Gigabit Net was co-funded by NSF/DARPA (91 -95); allowed limited high-speed services with industrial co-funding. • VBNS deliberately procured to private broader range advanced services – beyond commercial • Internet-2 put in for applications that need high speed and advanced services – donated fibre and other equipment • STARTAP set up for international entry to Internet-2 June 7, 2004 TNC 2004 20 Years of European Academic Networking 34

Advanced European Service Nets • Some NRENs have development portion – Surfnet has done it for a long time – RENATER, DFN and UKERNA has had it for some time – Has normally been a very limited resource • GEANT and some NRENs have special links into STARTAP – However they seldom make special arrangements inside their NRENs • 6 NET has been an exception for IPv 6 piloting – Has paved way for GEANT to be dual stack – Need for separate network now less, still only M/c, Qo. S • GEANT will make special arrangements for Grid Computing June 7, 2004 TNC 2004 20 Years of European Academic Networking 35

Advanced Services • New services, their use and their location drive network needs and deployment • Rise of web first widespread deployed WAN service – Required universal protocols – Would have ended protocol war if not nearly over in any case – Has led to unparalleled needs because of huge deployment and additions like image, streaming and real-time services June 7, 2004 TNC 2004 20 Years of European Academic Networking 36

Advanced Services on Nets • Adequate speed is necessary but not sufficient for many advanced services – – Videoconferencing has been supported by various EC and National projects for 12 years – Some variants require multicast – Voice/IP is coming on rapidly, and could be carried on the current networks – Streaming services available, but usage still limited • Better application support and Qo. S support still needed • Still tension between advanced and normal service – 6 NET spurred on GEANT; still no real peering June 7, 2004 TNC 2004 20 Years of European Academic Networking 37

Optical Networks • Emerging generation of optical networks raise difficult technical challenges • Offer immense possibilities for high speed applications • Allow wavelength partitioning of network to provide true isolation • With over-provisioning of international fibre, some is being donated – e. g. Gloriad • A new generation of world-wide optical interconnects is now coming into being June 7, 2004 TNC 2004 20 Years of European Academic Networking 39

Emerging Global Optical Network June 7, 2004 TNC 2004 20 Years of European Academic Networking 40

Drivers to Further Growth • Distance education, large-scale conferencing, tele-consultation and grid could be huge growth drivers • International growth could exceed local and national growth • Natural differences in data collections not yet utilised fully • Person-person real-time services hardly used so far • Widespread use of mobile access techniques – ADSL and 3 D/4 D cellular technology June 7, 2004 TNC 2004 20 Years of European Academic Networking 41

Constraints to Growth • Inadequate application support • Current base of charging and payment – E. g. distance education and telemedicine • Readiness to rely on remote computation – Central computing has beeen limited use • Lack of Distributed, easily usable, grid computing • Service limitations from security considerations • International Barriers – Language problems in foreign growth, sometimes worse – Intellectual Property Considerations – Natinalistic concerns – growth, pride, censorship June 7, 2004 TNC 2004 20 Years of European Academic Networking 42

Conclusions • The first decade brought large-scale backbones up to low Mbps, resolved the protocol wars, made remote communication standard and gave web access – Advanced islands had real-time services, gigabit communications, • The second decade brought backbones to low Gbps, remote access to Mbps, real-time communications readily available – Many remoter regions had the facilities of the previous decade in more advanced regions June 7, 2004 TNC 2004 20 Years of European Academic Networking 43

Conclusions • The move of GEANT to IPv 6 only fifteen months after 6 NETshowed academic BBs could move faster than earlier – Even that example shows that the move to advanced services still a tension • The next decade has technical capabilities almost in place to remove constraints of distance in academia – But major organisational, political and financial changes still needed to realise the full potential • Academic networks still have a role in advanced service experimentation and provision – Internet-2, WIDE and GEANT show this need well June 7, 2004 TNC 2004 20 Years of European Academic Networking 44

• The keynote for TNC 2014 will be very different • Electronic diffusion to the whole academic community should be a matter of course • But we can expect TNC 2014 to still be needed June 7, 2004 TNC 2004 20 Years of European Academic Networking 45