Internet2 Links and Glossary

To clarify the often cryptic acronyms and specialized terms associated with high-performance networking, here's a brief glossary with some links to web resources. . .

Networks/Organizations

• Internet 2 (I2) (http://www.internet2.edu): A new high-speed academic and research network connecting 135 leading higher education institutions directly or via a Gigapop. Gigapops and directly-connected I2 schools interconnect with each other via the vBNS or Abilene.
• vBNS (http://www.vbns.net): The Federal Very High Speed Backbone Network Service, operated by MCI for the NSF/ANIR. A high-speed academic and research network that predates Abilene. Most connections to it were at DS3 or OC3 speeds.
• Abilene (http://www.ucaid.edu/abilene): Another very high-speed academic and research network backbone, operated by Qwest in conjunction with Cisco and Nortel for UCAID. Connections to Abilene will be at OC3 or OC12 speeds. The UO will connect to it via two OC3 circuits. See http://boardwatch.internet.com/mag/98/jun/bwm61.html
• NSF/ANIR (http://www.cise.nsf.gov/ncri/index.html): The National Science Foundation's Advanced Networking Infrastructure and Research Division, which approves institutions for connections to the vBNS and related federally authorized high-performance academic and research networks like Abilene.
• UCAID (http://www.ucaid.edu/): University Corporation for Advanced Internet Development, formed by I2 largely to facilitate deployment of Abilene.
• MCI (http://www.governmentmci.com/mcigm/98/index.cfm): MCI Worldcom is a leading international telecommunication provider, formed via the merger of Worldcom and MCI in September 1998. Cable and Wireless (CWIX: http://www.cw-usa.net/) purchased most of MCI's Internet business in July 1998, excluding its vBNS and a few other governmental networking programs.
• Qwest (http://www.qwest.com/): Telecommunication company that operates the Abilene network. See http://www.wired.com/wired/archive/6.05/qwest.html?topic=connectivity&topic_set=newtechnology
• NGI (http://www.ngi.gov/): Next Generation Internet. Umbrella name for all Federal high-speed networking initiatives.

• NCO (http://www.ccic.gov/): National Coordination Office for Computing, Information and Communications. Coordinates Federal high-performance computing and networking initiatives.
• DARPA ITO (http://www.darpa.mil/ito/ResearchAreas.html): Defense Advanced Research Projects Agency Information Technology Office. Sponsors Supernet and many other high-performance networking and computing projects.
• Federal Mission Networks: Specialized, limited-access, high-speed networks deployed by individual Federal agencies in support of specific research programs or as a testbed for advanced technologies. Examples include

  - DREN (http://www.hpcm.dren.net/Htdocs/DREN/): Defense Research & Engineering Network

  - ESNet (http://www.es.net/): Department of Energy's Energy Sciences Net

  - NI (http://nic.nasa.gov/ni/): NASA Internet

  - NREN (http://www.nren.nasa.gov/): NASA Research & Engineering Network

  - Supernet (http://ale.east.isi.edu/NGI-S/): DARPA's Next Generation Internet

• Supercomputer Centers: Federally-funded centers that initially provided impetus for deploying many high-speed networking initiatives:

  - NCAR (http://www.ncar.ucar.edu/): National Center for Atmospheric Research

  - NCSA (http://www.ncsa.uiuc.edu/): National Computational Science Alliance

  - PSC (http://www.psc.edu/): Pittsburgh Supercomputer Center

  - SDSC (http://www.sdsc.edu/): San Diego Supercomputer Center. See also the National Partnership for Advanced Computational Infrastructure (http://www.npaci.edu/)

• NSF specialized networking support centers:

  - CAIDA (http://www.caida.org/): Cooperative Association for Internet Data Analysis

  - NLANR (http://www.nlanr.net/): National Laboratory for Applied Network Research. Charged with providing technical, engineering, and traffic analysis support of NSF High Performance Connections sites and HPNSP (high-performance network service providers) such as vBNS. NLANR projects include the development of Squid (http://squid.nlanr.net/), the leading web cache software product.

  - NSRC (http://www.nsrc.org/): Network Startup Resource Center, based at the UO. For the past decade, the NSRC has been involved in the deployment and integration of appropriate networking technology in various projects throughout the world.

  - StarTap (http://www.startap.net/): Science, Technology and Research Transit Access Point, located in Chicago, where approved foreign high-performance research networks connect to their American counterparts, such as the vBNS. Participating foreign networks currently include Canarie (Canada), SingAREN (Singapore), Transpac (APAN/Japan), MirNET (Russia), and TANet (Taiwan).

• ISOC (http://www.isoc.org/): A professional membership society with over 100 organizational and 6,000 individual members in 100+ countries. A leader in addressing issues confronting the future of the Internet and a home to the groups responsible for Internet infrastructure standards, including the Internet Engineering Task Force (IETF) and the Internet Architecture Board (IAB).
• IETF (http://www.ietf.org/): A large international community of network designers, operators, vendors, and researchers concerned with the evolution and smooth operation of the Internet. Open to any individual.

• NANOG (http://www.nanog.org): North American Network Operators Group. An association of network service providers and others interested in operational details of national networks.
• Wide area circuit speeds

  - DS0: 64 kilobits a second (about the speed of a fast dialup modem)

  - T1: 1.544 megabits a second (typical speed of a small/mid-sized ISP's connection to the net)

  - DS3: 45 megabits a second (the speed of OSU's vBNS connection)

  - OC3: 155 megabits a second (the UO will be connected to Abilene via two OC3 circuits)

  - OC12: 622 megabits a second (the speed of the highest speed national commodity network backbones)

  - OC48: 2.4 gigabits a second (the Abilene backbone speed at startup)

  - OC192: 9.6 gigabits a second (eventual speed of the Abilene backbone)

  - ethernet: 10 megabits/second (the speed of most UOnet connections)

  - fast ethernet: 100 megabits/second (the speed at which many campus servers connect to UOnet. Where available, individual users can purchase fast ethernet connects for $250 each)

  - gigabit ethernet: 1000 megabits/second (current speed of UOnet backbone)

• Shared/switched ethernet:

  - shared ethernet: traditionally, ethernet has been a shared media, with all users on a subnet sharing the bandwidth available on that circuit. Cost-effective, but a poor choice for high-performance connections.

  - switched ethernet: Connections that dedicate the full capacity of a connection to a particular workstation or server, and limit the traffic sent to a particular host to traffic that is destined for, or coming from, that host. Switched ethernet is normally how high-performance systems will connect to UOnet.

• Half duplex/full duplex:

  - half duplex: on a half duplex connection, a connection talks or listens, but doesn't do both at the same time.

  - full duplex: on a full duplex connection, a workstation or server can simultaneously transmit and receive data.

• FDDI: an older 100 megabit-per-second competitor to fast ethernet. See a comparison of FDDI and Fast Ethernet at http://www.cisco.com/warp/public/729/c5000/swfet_wp.htm
• ATM: Asynchronous Transfer Mode. Circuit-switched (rather than packet-switched) network technology that chops data into 53 byte cells. Often used for wide-area high-performance connectivity. For an excellent ATM tutorial, see http://www.scan-technologies.com/tutorials/ATM%20Tutorial.htm
• SONET (and Packet Over SONET): The alternative to ATM for wide-area, high-performance connectivity via Abilene, and the option the UO will be using. Quoting Cisco¹s Packet Over SONET whitepaper: "For expensive WAN links, Packet over SONET can provide as much as 25- to 30-percent higher throughput than Asynchronous Transfer Mode (ATM)-based networks. By transporting frames directly into the SONET/SDH payload, the overhead required in ATM cell header, IP over ATM encapsulation, and segmentation and reassembly (SAR) functionality is eliminated." For the complete discussion, see http://www.cisco.com/warp/public/733/12000/gspos_an.htm A general SONET tutorial is available at http://www.webproforum.com/tektronix/index.html
• WDM/DWDM: Wave Division Multiplexing/Dense Wave Division Multiplexing. The underlying technology allowing carriers with existing fiber to obtain huge amounts of incremental bandwidth without having to pull additional fiber. Allows a single strand of fiber to simultaneously carry multiple parallel transmissions on different light wavelengths ("lambdas"). A good WDM tutorial is available at http://www.webproforum.com/lucent3/
• GSR: Cisco's 12000-series Gigabit Switch Router, the special high-performance router that many sites, including the UO, are using to connect to Abilene. See http://www.cisco.com/warp/public/733/12000/

Network Architecture and Operations

• Gigapop: Gigabit point-of-presence. As originally envisioned, I2 sites would connect to the I2 backbone via a regional "Gigapop." Gigapops would combine the traffic from directly connected I2 member sites and exchange it with other Gigapops via I2 backbone links. By aggregating individual I2 sites behind Gigapops, the size of the I2 routing matrices can be kept small and efficient, and network problems could be resolved via contact between Gigapop NOCs. The Oregon Gigapop is located in Eugene and operated by the UO.
• POP (Point-of-Presence): Location at which an internet service provider connects to the Internet and offers local connections to customers, whether via leased circuits or dialup connections. A small ISP might have only one POP, while a national ISP might have a POP in virtually every major metro area.
• Exchange point (also, "meet point"): A location where network service providers can peer and exchange network traffic with each other. The UO operates the Oregon Internet Exchange in Eugene (see http://antc.uoregon.edu/OREGON-EXCHANGE/ )
• Peering/Transit: When two networks peer, they agree to cooperatively exchange their network customers' traffic--and only their customers' traffic--typically at no charge. These traffic exchanges take place at an exchange point. Contrast peering with buying transit bandwidth from a network service provider: when you buy transit bandwidth, the network service provider agrees to accept and transfer traffic on your behalf to any other destination on the Internet, whether it's for another customer of the NSP's or for a customer of some other network service provider.
• Meritorious/Commodity Traffic: Meritorious traffic meets the Acceptable Use Policy (AUP) adopted by a particular network. For example, I2's AUP limits use of I2 to traffic destined for other I2 sites or comparable/approved research and education networks. Other traffic, such as a mail message to an AOL user or a web page from CNN's web site, would be considered "commodity traffic" and would not be eligible to travel over I2.
• NOC: Network Operation Center. A NOC is a facility, typically staffed 24 hours a day/7 days a week, by NOC technicians who can monitor, detect and arrange for a response to network problems. There's usually one NOC per network.
• Colocation: While most networked companies or schools want to have machines in their facilities connected to the Internet, some companies don't really care where their servers are physically located. For example, a company¹s main web server might just as easily live in a locked rack at an exchange point as in a company office. Servers located at network service provider facilities rather than at corporate or campus locations are said to be "colocated" and are typically able to buy bandwidth for less than the cost of a traditional Internet connection due to the elimination of local loop charges and reduced equipment requirements for the network service provider. Colocation also typically offers the server owner access to conditioned/uninterruptable power, improved physical security, geographical flexibility in siting servers, etc.
• Local Loop: A synonym for "local access circuits" or the physical wire or fiber optic cable that connects a site to their Network Service Provider's POP. Local loops are typically purchased from the local phone company or from a competitive local exchange carrier. For example, the UO buys its local loops from USWest and Electric Lightwave (ELI). Local loop charges are in addition to a network service provider's port charges.
• Port Charge: The payment made to a network service provider so they will accept and transfer your transit traffic. Port charges are normally sized according to the maximum capacity of the circuit to the network service provider.
• Backhaul: If a network service provider doesn¹t have sufficient capacity (or required services) at a local POP, they may need to "backhaul" a circuit you¹ve ordered to another better-equipped location, which adds to the cost of obtaining that service (unless backhaul charges are waived).
• IPv4/IPv6: IP (also, ³IPv4²) is the Internet Protocol that underlies virtually all Internet traffic today. A new version of the IP protocol, IPv6, is on experimental use on the so-called 6Bone (see http://www.6bone.net/). IPv6 access is an example of the sort of advanced services that Internet 2 will likely support. See the tutorial at http://www.6bone.net/case-for-ipv6.txt or see http://playground.sun.com/pub/ipng/html/INET-IPng-Paper.html
• IP multicast video: With IP multicast, a single network video stream services all viewers on a given network segment, rather than requiring separate streams for each viewer. IP multicast scales well to virtually unlimited-sized global audiences. Cisco¹s excellent IP multicast training materials are available online at ftp://ftpeng.cisco.com/ipmulticast/ multicast_training.html An easy way to begin using IP multicast is to install Cisco/Precept¹s IP/TV from the UO Duckware CD-ROM, or see http://darkwing.uoregon.edu/~joelja/project/mbone.html
• QoS (Quality of Service): There has been a lot of interest among I2 participants in Quality of Service-related issues, or efforts to provide better than ³best-efforts² access to Internet resources. For example, some Internet applications (such as tele-surgery or remote management of a scientific instrument) require bounds on latency and jitter, or a reserved amount of bandwidth from end-to-end. An excellent series of papers are available at http://www.internet2.edu/qos/may98Workshop/html/presentations.html
• Bandwidth and Latency: Bandwidth can be considered the effective rate at which bits are delivered from their source to their destination‹or the "size of the pipe" between two sites. Latency is the delay a packet experiences while being transmitted from its origin to its destination. This is a crucial issue in high-performance networking because there is a close relationship between packet latency and the maximum effective bandwidth which a connection can sustain. See http://engr.ans.net/slides.html#nanog-feb-1997-tcp-congested
• Jitter: The variance in the time it takes for a packet to be transmitted from its origin to its destination. Jitter is a problem in advanced applications such as IP multicast video because even small deviations in delivery pacing can translate into noticeable interruptions in otherwise smooth video delivery, unless buffering or other similar techniques are employed.