Welkom bij SURFnet SNB-master Dennis Paus, Roeland Nuijts, Erik-Jan Bos SNB-master 2004/2005 Utrecht, 23 maart 2005

AgendaAgenda 10:00 uur:Ontvangst met koffie en thee 10:15 uur:Welkom en Opening (Erik-Jan) 10:20 uur:"SURFnet5, ins and outs" (Dennis) 11:00 uur:"GigaPort en SURFnet6" (Erik-Jan) 11:45 uur:Bio-break 12:00 uur: "In-depth optische technologie" (Roeland) 12:45 uur:Vragen 13:00 uur:Lunch

SURFnetSURFnet  Provides the Dutch National Research Network  Not for profit company, 55 employees  100% subsidiary of Stichting SURF  170 connected organizations, users  Turnover (2003): 30 M€  Infrastructure services: –innovation paid for by government –cost effective exploitation for higher education and research

GigaPort Next Generation Network  Research networking as innovation engine between research and market introduction of new services GigaPort NG Network ( )  Consortium of 50 Dutch research organizations  Government grant 40 M€  Project started January 1 st, 2004  Builds on GigaPort project ended in 2003  Partnership with industry

AgendaAgenda 10:00 uur:Ontvangst met koffie en thee 10:15 uur:Welkom en Opening (Erik-Jan) 10:20 uur:"SURFnet5, ins and outs" (Dennis) 11:00 uur:"GigaPort en SURFnet6" (Erik-Jan) 11:45 uur:Bio-break 12:00 uur: "In-depth optische technologie" (Roeland) 12:45 uur:Vragen 13:00 uur:Lunch

GigaPort en SURFnet6 Erik-Jan Bos Director of Network Services, SURFnet SNB-master 2004/2005 Utrecht, 23 maart 2005

History of the SURFnet infrastructure 10 kbit/s 100 kbit/s 1 Mbit/s 10 Mbit/s 100 Mbit/s 1 Gbit/s 10 Gbit/s 100 Gbit/s SURFnet1 9,6 kbit/s SURFnet2 64 kbit/s SURFnet3 2 Mbit/s SURFnet4 34 Mbit/s SURFnet4 155 Mbit/s SURFnet5 20 Gbit/s

SURFnet continuing paradigm shifts SURFnet1 Public X.25 ==> SURFnet2 Private X.25 SURFnet2 X.25 ==> SURFnet3 IP SURFnet4 IP over ATM SURFnet4 IP over SDH SURFnet5 IP over lambdas

Paradigm shift SURFnet4 projectGigaPort DWDM Lambdas POS 1995 GigaPort Next Generation SURFnet4 network SURFnet5 network ATM 2008 SURFnet6 network Next generation is not a simple extrapolation of current networks

IP Traffic growth, connected organizations 1600 Tbyte/maand ≈ 1 CD-ROM/seconde

SURFnet5 lessons learned Dark fiber is the essential building block for next gen networks Lambdas form an excellent basis for IP networking Researchers are interested in lambdas Provides excellent quality on point-to-point connections at very high speed Protects the routed network Enables demanding applications to make use of the infrastructure in an economically sound way

A word on networking costs Costs of optical port is 10% of switching port is 10% of router port with same characteristics –10G routerblade -> k$, 10G switch port -> k$, MEMS port -> 0.7 k$ Give each packet in the network the service it needs, but no more This slide courtesy of Cees de Laat (UvA)

Lesson: Transport in metal and glass Electrons in a metal wire (e.g. Cu): –Spin ½, Fermi-Dirac statistics (Fermions) –Collisions: resistance and hampered “flow” –Very suitable for light bulbs and toasters Fotons in a glass wire (fiber): –Spin 1, Bose-Einstein statistics (Bosons) –No collisions, zero resistance and unlimited flow –“Superconductors” for data

Laying of fiber near/at Science Park Amsterdam Pictures by Yuri Demchenko

VLBI at JIVE in Dwingeloo, NL today

eEVN: European VLBI Network 1-30 Gbps Data processing centre: 16 Gbps (2005) 1 Tbps (2010) China USA South Africa Russia asymmetric star topology This slide courtesy of Richard Schilizzi

eInfrastructures Workshop – Den Haag CMV/2004/11/18 LOFAR as a Sensor Network LOFAR is a large distributed research infrastructure: Astronomy: –>100 phased array stations –Combined in aperture synthesis array –13,000 small “LF” antennas –13,000 small “HF” tiles Geophysics: –18 vibration sensors per station –Infrasound detector per station >20 Tbit/s generated digitally >40 Tflop/s supercomputer innovative software systems –new calibration approaches –full distributed control –VO and Grid integration –datamining and visualisation Slide courtesy of Marco de Vos (LOFAR)

SURFnet6 overview  A hybrid optical and packet switching infrastructure  Based on customer-owned managed dark fiber  Native IPv4, IPv6 and Light Path Provisioning over a single transmission infrastructure –Managed via a single control plane –Network nodes reduced from 20 routed locations to 2 routed locations Paving the way to a ubiquitous and scalable Services Grid

Timelines SURFnet6

SURFnet’s new Industry Partners ( )  Leader of the consortium  Optical equipment  Ethernet equipment  Network management equipment  Routing equipment  Installation services  Maintenance services

SURFnet6 on dark fiber  SURFnet6 will be entirely based on SURFnet owned managed dark fiber via the customer premises  Over 5300 km fiber pairs available today; average price paid for 15 year IRUs: < 6 € /meter per pair  Managed dark fiber infrastructure will be extended with new routes, to be ready for SURFnet6

Common Photonic Layer (CPL) in SURFnet6

SURFnet6: IP Services  IPv4 and IPv6 connectivity –Unicast –Multicast  1 and 10 Gigabit Ethernet connections  Small routed IP core in Amsterdam at two separate locations  Congestion-free via overprovisioning  Resilient

SURFnet6: IP network implementation Avici SSR External IP connectivity SURFnet6 Core Routers SURFnet6 Border Routers SURFnet6 Common Photonic Layer 10 Gigabit Ethernet Customer Avici SSR Avici SSR Avici SSR Non-SURFnet SURFnet infrastructure Nortel Passport GE Nortel OM GE CPE Nortel OME Gigabit Ethernet Customer CPE 1 GE Nortel OME 6500 Nortel OM 5000 Nortel OM 5000 Nortel OME 6500 Nortel OME 6500 Nortel OME 6500

SURFnet6: Light Path Provisioning Lambdas:  enable layer 1/2 end-to-end Light Paths Light paths:  provide excellent quality on point-to-point connections at very high speed (1-10G)  not constrained by traditional framing, routing, and transport protocols  are becoming integral part of scientific instruments  enable creation of Optical Private Networks (OPN)

SURFnet6: Light Path Provisioning implementation GLIF SURFnet6 Common Photonic Layer Customer equipment Non-SURFnet SURFnet infrastructure Optical Switch 10 GE 1 GE 10 GE Customer equipment End-to-End Light Path 10 GE LAN Amsterdam International Light Paths 10 GE Nortel OME 6500 Nortel OME 6500 Nortel OME 6500 Nortel OME x16 MEMS 16x16 MEMS Nortel HDXc

Photonic XC enables dynamic lambda networking A patch panel, like a telephone switchboard… Control plane … but automated

Groningen1 PoP infrastructure

Optical Private Network (OPN)

NetherLight: Open Optical Exchange  Experiments with Light Path provisioning in a multi-domain environment, successful demo at SC04 with Nortel and UvA  Open Optical Exchange in Amsterdam –Operational since January 2002 –Built and operated by SURFnet  Nortel Networks HDXc at the centre  Full duplex 640G non-blocking cross- connect capability  GE grooming, GE switch for access to clusters

NetherLight Q DWDM SURFnet 10 Gbit/s SURFnet 10 Gbit/s SURFnet 10 Gbit/s Dwingeloo ASTRON/JIVE Dwingeloo ASTRON/JIVE Prague CzechLight Prague CzechLight 2.5 Gbit/s NSF 10 Gbit/s London UKLight London UKLight Stockholm NorthernLight Stockholm NorthernLight CESNET 10 Gbit/s UKERNA 10 Gbit/s Geneva CERN Geneva CERN Science Park Amsterdam Chicago IEEAF 10 Gbit/s SURFnet 10 Gbit/s New York MANLAN SURFnet 10 Gbit/s

GLIF: Global Lambda Integrated Facility  Established at the 3 rd LambdaGrid Workshop, August 2003 in Reykjavik, Iceland  Collaborative initiative among worldwide NRENs, institutions and their users  A world-scale Lambda-based Laboratory for application and middleware development GLIF vision : To build a new grid-computing paradigm, in which the central architectural element is optical networks, not computers, to support this decade’s most demanding E-science applications.

GLIF after Nottingham Helps to develop the Global LambdaGrid for the advancement of scientific collaboration and discovery GLIF “glues” together the networks and resources of its participants GLIF is not a new network and GLIF will not compete with its participants GLIF will be positioned on the demand site of the market and managed as a cooperative activity TERENA will act as the GLIF Secretariat

GLIF World Map – December 2004 Visualization courtesy of Bob Patterson, NCSA.

ConclusionConclusion  Users need new services that current networks cannot support  Telecommunication infrastructures will become part of the Grid and will be integrated in scientific instruments  Hybrid networks delivering IP and Lambda Services can meet user demand within budget constraints  SURFnet6 will enable advanced multimedia collaboration and applications, and the use of Grid technology in research

Thank you