National LightRail: Towards a National Optical Research Network Marla Meehl UMC 2/20/03

2 National Light Rail Dark Fiber National footprint –Building from California via Denver to Atlanta initially and completing national footprint in a ring back to California as funds and partners available Serves very high-end experimental and research applications Gb Wavelengths initially Capable of 40 10Gbps wavelengths at build-out Partnership model

NLR Footprint and Layer 1 Topology Terminal, Regen or OADM site (OpAmp sites not shown) Fiber route PIT POR FRE RAL WAL NAS PHO OLG ATL CHI CLE KAN OGD SAC BOS NYC WDC STR DAL DEN LAX SVL SEA SDG JAC

4 Goal of NLR “The fundamental and overriding goal of NLR is to provide an enabling experimental infrastructure for new forms and methods of science and engineering.” – Peter O’Neil

5 Elements of NLR Enable science that cannot be done today Distributed Terascale Facility/Extended Terascaled Facility (DTF/ETF) access Grid/Distributed computing Distributed Storage Area Networks (SANs) Real-time, high-speed data access and service Distributed coupled modeling

6 Elements of NLR There are three broad areas of research activity that the proposed NLR infrastructure would facilitate: – Experimental Computational Science and Engineering Application Research grid computing, load migration, latency tolerant algorithms, resource aware algorithm selection and scheduling, distributed and multidisciplinary efforts involving multiple length and time scales in the simulations and multiple science and engineering disciplines, very tightly-coupled simulation codes and algorithms, data sensing and collection elements, data repositories, visualization environments, etc.

7 Elements of NLR High Performance Networking Research conduct network layer or link-layer research on some wavelengths, while maintaining a stable link layer and network layer on other wavelengths to conduct research at the transport layer or higher layers build the next generation of network protocols using a protocol development environment (PDE) conduct research activities in the optronics space itself

Research and Development Commercialization Partnerships Privatization Today’s Internet ‘NGI’ efforts Source: Ivan Moura Campos Research Nets Evolution of Networking ‘Spiral’ And, Long Term

Common Denominator, Advanced Services Research & Education Network Bleeding-edge e2e services & facilities for key experiments, programs, projects, researchers & for next gen. tech., architectures, Grids, content, apparatus, etc. ‘ Things that haven’t been done before; with, to, on or over networks’. Leading-Edge ip Services for demanding apps. & middleware Advanced Services for R&E users Business,.Gov. & General Public Commodity Internet I2-Abilene, GigaPoPs; Fed. ‘NGI’ Nets. Internet-2, GigaPoPs I2-K20 ISP’s, hosting.com’s etc. Operational High Performance Research Support Networks CS, Comp. Science & ‘Grid’ researchers, + major experiments (eg DTF ) & net. experimenters + leading edge labs and centers, optics sys.& net. Industry R&D groups + I-WIRE P ACIFIC L IGHT R AIL + Internet-2 wave projects ; Research ’s, P2p fiber, & >=10g ip … Experimental & Developmental Networks LEADERS NETWORK TYPE CAPABILITIES/USERS R&E Network Tiers NETWORK DISCONTINUITIES 12/05/01 Computer Science & Net. Researchers Academic Research, papers, lab. Experiments, spools of fiber In labs & sparse testbeds

10 Drivers and Motivations “NSF must be prepared to assume a great S&E infrastructure role for the benefit of the Nation.” An increasing number of researcher and educators, working as individuals and in groups, need to be connected to sophisticated array of facilities, instruments, and data bases.” Science and Engineering Infrastructure for the 21st Century Report

11 Drivers and Motivations “Develop and deploy an advanced cyberinfrastructure to enable new Science & Engineering in the 21st century” “Expand education and training opportunities at new and existing research facilities.” “Develop interagency plans and strategies to provide high-end high performance computing and networking infrastructure.” Science and Engineering Infrastructure for the 21st Century Report

12 Drivers and Motivations Cost Control Flexibility Responsiveness Innovation End-to-End networking

Rationale for NLR: Current market demonstrates willingness of private sector setting stage for strong partnership of academia, corporations and government Research community in USA and worldwide working on large scale research and development applications calling for increasing bandwidth as evidenced by the Grid and Terascale projects Unified approach best investment for biggest payoff for the most users Federal program funding support focused on research and experimental applications needing significant network capacity Leadership worldwide in research and education increasingly dependent upon large-scale collaborative research activities Meet the future needs of the research and community most economically

14 Why Fiber? Capacity needed is not otherwise affordable Capabilities needed are not available Cheaper in the long range Leverage with “carriers” Insurance against monopoly behavior Stable and predictable anchor points

15 NLR Service Offerings Waves!!!! Dark Fiber ‘MetaPoP’ capabilities Dedicated OC-192c or 10GE waves –No wave can interfere with others Shared IP service –Must be able to tolerate outages –Supports experimental code Shared Layer 2 server Access to 2 unused waves Subdivided waves for dedicated uses –Delivered as gigabit Ethernet

16 NLR Governance LLC - Equity Partners Joint Research Council Joint Technology Council Distributed Network Engineering/ Operations Waves governed/managed by various groups

17 Potential Participants Carnegie Mellon/PSC Virginia Tech Duke Georgia Tech Florida Group (FSU lead) Texas Group (UT Austin lead) CENIC NOTE: There are many sites behind each of these lead sites PNWGP UCAR UCAID

18 Technology Partners Cisco Systems Level 3

19 Total NLR Costs $80.2 Million Over 5 Years

20 Westnet LightRail (WLR) Costs $4.6 Million over 5 years Westnet LightRail is a consortium of the FRGP Abilene members and the InterMountain GigaPop (IMG) –University of Utah, Utah State, Utah Education Network Share geography and research goals IMG would become members of the FRGP to help off-set total costs

21 Costs to UCAR/UCAR Share commodity Internet, engineer, equipment, Abilene, and NLR costs $100-$200K new dollars per year for UCAR for five years depending on WLR financial commitments Without WLR consortium the cost to UCAR would be $920K per year for five years

22 UCAR Benefits/Risks Benefits Enables advanced scientific research Least expensive way to obtain DTF/ETF waves to LA or Chicago Positions participation in NSF solicitations e.g. Experimental Infrastructure Network and Networking Research Testbeds Provides experimental lambda to PSC for jointly funded work Further aggregation of Abilene and commodity links to reduce future costs Founding partner least cost/ potential profit position Risks As founding partner, UCAR provider of services for WLR members - bears more financial & liability risks FRGP and IMG may take different time frames to gather funds WLR members can’t join corporations of any kind Puts 5 year liability w/o offset asset on our books Some risk, but still most prudent and cost effective way for UCAR to proceed Liability and financial risks are being further evaluated by legal and finance

Questions?