Recent Development in Networking Connectivity Michael Ernst DESY

Recent Development in Networking Connectivity Michael Ernst DESY 12/08/01 Ernst 1

DFN’s Trans Atlantic Connectivity Today TM 4 S STM 4 12/08/01 Ernst 2

News on TA Connectivity (1) • Planned for 2002 – Direct link GEANT <-> Abilene / CANARIE @ 2 * 2. 5 Gbps (Q I, 2002) – UCAID will add another 2 * 2. 5 Gbps – UCAID proposing joint Project GTRN 12/08/01 Ernst 3

DFN’s Trans Atlantic Connectivity in Q I, 2002 M ST 16 STM 4 STM 16 12/08/01 Ernst 4

News on TA Connectivity (2) • Global Terabit Research Network – “Terabit Net” for Science and Research – Europe as Mediator for Asian-Pacific Area • For DFN/US Global Transit (commodity traffic) – 2 * 2. 5 Gbps in Q I, 2002 – Contract with 2 Providers • Global Crossing (2. 5 Gbps) • KPNQwest (2. 5 Gbps) – Tendering Process in Collaboration with DANTE 12/08/01 Ernst 5

Trans Atlantic Traffic Development from Sep ‘ 00 - Nov ‘ 01 (Statistics provided by DFN) 12/08/01 Ernst 6

GÉANT Gigabit Speeds • Backbone speeds: – initial target: core at 2. 5 Gbps – network achieved: 9 trunks at 10 Gbps and 11 trunks at 2. 5 Gbps • Access speeds: – 11 NRENs to connect at 2. 5 Gbps • Future plans: – 100’s of Gbps within four years 12/08/01 Ernst 7

GÉANT Geographic Expansion • TEN-155: – 21 NRENs linked – 25 countries connected in total • GÉANT: – 27 NRENS – 31 countries - 6 new countries, 2 in the Balkans – Open to further connections from national organisations 12/08/01 Ernst 8

GÉANT Global Connectivity • Connectivity to other regions of the world • European Distributed Access: – effective distribution mechanism through backbone core – uniform presentation of the European networks 12/08/01 Ernst 9

GÉANT Guaranteed Qo. S • TEN-155 continuity: – TEN-155 Managed Bandwidth Service, using ATM technology • End-to-end Qo. S – with NREN involvement • Different types of Qo. S: – Guaranteed bandwidth – Predictable delay and jitter – Guaranteed bandwidth + predictable delay and jitter 12/08/01 Ernst 10

GÉANT - A Development Platform • A network for research: – TF-NGN, SEQUIN • http: //www. dante. net/tf-ngn • http: //www. dante. net/sequin • Initiatives with IPv 6: – TF-NGN, 6 NET Proposal • Co-operation with Grids: – Euro. Grid, Data. Grid • Joint development clauses with suppliers 12/08/01 Ernst 11

Price Development of International Bandwidth (Source: V. Berkhout) 12/08/01 Ernst 12

GÉANT Technology • 2. 5/10 Gbps as single wavelengths – with SDH framing – no access to optical level / (dark) fibre • TEN-155 aggregate circuit length = 22, 000 km 12/08/01 Ernst 13

Stepping to 10 Gbps • Cost vs capacity – 10 Gbps in 8 countries … • quadruples capacity • adds less than 10% to cost – Pushing 2. 5 Gbps as far as possible 12/08/01 Ernst 14

Issues • Service is diverse (wavelengths and SDH) • Guaranteed bandwidth and VPN capability needs further development. . . • . . . as does management of end-to-end capability • Cooperation of NREN, MAN and campus network operators is needed 12/08/01 Ernst 15

Summary on GEANT • 10 Gbps now! • Wavelength networks – National and International • • 12/08/01 Better geographical coverage Progress on global connectivity Challenging the vendors Qo. S/VPN will need our help! Ernst 16

DFN’s Connectivity to Europe GEANT STM 16 (2. 5 Gbps) 12/08/01 Ernst 17

Development of DFN <-> TEN-155 / GEANT Traffic from Jan ‘ 01 - Nov ‘ 01 (Statistics provided by DFN) 12/08/01 Ernst 18

DESY’s Monthly Traffic Volume (Received) GB 12/08/01 Ernst 19

The Silk Project Hans Frese DESY Hamburg NATO Advisory Panel on Computer Networking SILK Task Force 12/08/01 Ernst 20

The Silk Task Force • • Peter Kirstein, UCL, London Hans Frese, DESY, Hamburg Robert Janz, RUG, Groningen Sergey Berezhnev, MSU, Moscow Zita Wenzel, ISI, Marina del Rey, CA Rolf Nordhagen, UIO, Oslo Ruben Mkrtchyan, YERPHI, Yerevan Walter Kaffenberger, NATO 12/08/01 Ernst 21

Activity of the Network Panel • The Networks Panel has supported Network Infrastructure Grants (NIGs) for many years – Was initially Russia and Eastern Europe – Southern Caucasus and Central Asia are current principal areas for our larger grants • Internet Connectivity has been a large part of each NIG • Current bandwidths much too small – but all that can be afforded from budget 12/08/01 Ernst 22

Intentions of NIGs • Improve National Research Net Infrastructure – Not that of isolated groups or institutes • Encourage National Collaboration – Preferably to set up National Research and Education Networks (NRENs) • Encourage International Collaboration – Ever more important at the current time 12/08/01 Ernst 23

Current Connectivity • Bandwidth from NATO sources currently 64 – 512 Kbps – Would like to go up by an order of magnitude at least – Cost unaffordable in current model ($100 k per year for 1 Megabit per second) • National Research and Education Networks (NRENs) partially exist in most of the countries intended currently 12/08/01 Ernst 24

Possible Technologies • Mainly Fibre in Western Europe – No affordable fibre yet in Caucasus or Central Asia (> 5 times satellite cost) – Does exist in E. Europe and Russia • Satellite attractive in these areas – Satellite Bandwidth driving force – Broadcast capability can be useful • Proposed Silk Project in 2000 – Based on VSAT Technology 12/08/01 Ernst 25

Schematic of the Silk System 12/08/01 Ernst 26

A short primer on satellites (1) • Satellites are bent pipes in the sky • 5 to 10 year lead times imply mature/old technology in the sky • Fiber has taken over the oceans, satellites are looking for work • Compared to fiber, satellite bandwidth is low, but. . . 12/08/01 Ernst 27

A short primer on satellites (2) • . . . one satellite covers one third of the globe • . . . broadcasting to many locations is trivial • . . . bandwidth is simplex – it can be allocated asymmetrically and shared between locations • . . . minimal local infrastructure requirements – unobstructed view south – 240 Volts with diesel backup if needed 12/08/01 Ernst 28

A short primer on satellites (3) • Buying satellite bandwidth: – you pay for radio frequency bandwidth – and battery consumption in the sky – buy in bulk to obtain discount • Using a larger dish fetches more energy and improves the signal to noise ratio • This allows higher density modulation which produces more Mbps per MHz • Net result: 1 Mbps per year for $25 K 12/08/01 Ernst 29

Who gets connected? • Funded by NATO – National Research and Education Networks (NRENs) in the Partner countries • Co-funded by – NGOs – Supranational Organisations • Staged Implementation – E. g. Armenia, Georgia, Kyrgyz Rep, Uzbekistan – E. Azerbaijan, Kazakhstan, Tajikistan, Turkmenistan 12/08/01 Ernst 30

Satellite station configuration 12/08/01 Ernst 31

Planned Silk Bandwidth 12/08/01 Ernst 32

Eurasia. Sat • PROs • CONs – Covers both Asia and Western Europe – Can use small earth stations (2. 4 m) – Hamburg hub (DESY) well connected to European backbone – Bandwidth on demand for e. g. teleteaching 12/08/01 Ernst – Requires expensive hub station in the West ($650 K) 33

Project Management • DESY will provide Technical Management • Area Consultants plus Silk Task Force will provide first Process Management – If project grows, may get professionals – May be able to work with other funding agencies operating in the area • Project Steering Committee provide high level policy management 12/08/01 Ernst 34

Policy Steering Committee • Will include at least all Co-Directors and representatives of funders • Terms of Reference to be decided by members • Policy includes many areas to discuss – AUP rules, Membership, responsibilities of NRENs, move to financial sustainability, bandwidth rules, any dispute resolution 12/08/01 Ernst 35

Summary • Propose a system with 25 -50 Mbps for 8 countries in Caucasus and Central Asia • System could grow with additional investment from others 12/08/01 Ernst 36

Conclusions • The Project is ready to start • The technical and organisational framework is in place • A choice has to be made 12/08/01 Ernst 37

Modular Campus Network Design Building Access Layer L 2 Switch L 3 Switch = Router Distribution Layer Backbone L 3 Switch = Router Server Distribution L 2 Switch Server Farm 12/08/01 Ernst 38