BOLD 2003: Development and the Internet

PART 3 - Alphabet Soup (or "Who Are All These Mysterious Internet Elves, Anyway?")

Let's take another look at Andrew and Ethan's email exchange, from an organizational perspective. Who are the key players that designed, deployed and/or now operate the Internet? We don't have time to go into everyone in detail, but it's enlightening -- and sometimes surprising -- to have a brief introduction to the names and faces.

• The Architects: Standards bodies

Importance. The Internet is built on open standards: technical specification documents that are published by technical standards bodies. What is really interesting and unprecedented -- some have said revolutionary -- about the Internet is that its key standards have been, and are still, developed through open processes, by teams of technical experts who work together to identify, study, and solve common engineering problems. After an exhaustive process of peer review within the community of Internet engineers and architects (including, essentially, anyone who chooses to participate and contribute), the resulting documents -- the standards themselves -- are published openly, and for free. At that point, it is up to the network operators and software writers to use or not use a newly-published standard. Because the Internet is a global network of voluntarily interconnected networks, there is no organization that can force it to adopt a new standard or technical protocol.

History. Initially, Internet standards were produced by researchers working for, or funded by, the Defense Advanced Research Projects Agency (DARPA), the US Department of Defense's main research and development arm. In 1969, DARPA commissioned the first packet-switched wide-area network (and the first direct ancestor of today's Internet), the ARPANET. The TCP/IP protocol that we discussed in Part 2 of this lecture was developed and deployed in the 1970s by DAPRA researchers and teams at Stanford University, UCLA, USC, the University of California - Santa Barbara, the University of Utah, the University of California - Berkeley, MIT, Stanford Research Institute, and BBN, a pioneering DARPA contractor. During the 1980s, new players built their own packet-switched networks, using TCP/IP and other protocols developed for the ARPANET, and developing new standards. These included:

• the US National Science Foundation (CSNET and NSFNET)
• AT&T Bell Laboratories (the UNIX operating system, which led to the USENET)
• various United States universities (BITNET)
• a consortium of the University of Michigan's MERIT organization, IBM, and MCI (NSFNET)
• NASA (SPAN)
• the US Department of Energy (MFENet and HEPNet)

Often, these networks were deployed using independent standards and approaches, but all eventually migrated to the TCP/IP protocol suite. Many other organizations and institutions played roles in the early development and implementation of the Internet's protocols; for a solid historical overview by some of the founding fathers of the Internet, see "A Brief History of the Internet."

The early 1990s saw the transformation of the Internet from an academic and research network to a privately-owned network of networks, worldwide. In 1992, Congress authorized the opening of the NSFNET to commercial use. At the same time, the center of gravity for the development and refinement of Internet standards and protocols shifted from the academic experts, working under government research grants, to new institutional settings, populated overwhelmingly by network engineers and architects affiliated with commercial businesses.

IETF. Today, the premier standards body for the Internet is the Internet Engineering Task Force (IETF). The IETF describes itself as "a large open international community of network designers, operators, vendors, and researchers concerned with the evolution of the Internet architecture and the smooth operation of the Internet." The IETF has no "membership" as such -- anyone who is interested can participate by joining a working group mailing list or attending one of the thrice-yearly face-to-face IETF meetings. The IETF's standards process is managed by the Internet Engineering Steering Group (IESG). Overall architectural oversight and guidance is provided by the Internet Architecture Board (IAB). The IAB and IESG are chartered by the Internet Society (ISOC), which is a global membership society for Internet professionals (by the way, membership in ISOC is free -- join here!). The IETF, IAB, and IESG work on standards and protocols that run over the Internet protocol. The IETF, IAB, and IESG are virtual organizations; they have no headquarters or physical offices. The ISOC is headquartered in Reston, Virginia, and Geneva, Switzerland. The output of the IETF is published in the RFC series. See [Note 3].

W3C. One of the most popular applications that runs over the Internet protocol is the World Wide Web. The World Wide Web Consortium (W3C) is the key standards body for the Web, developing and publishing protocols and standards. Unlike the IETF, the W3C is a membership organization with about 450 organizational members, and utilizes a dedicated professional staff of around 70 who contribute to the development of W3C specifications (and related software tools). The W3C is hosted by three institutions on three continents: the Massachusetts Institute of Technology (MIT) in the United States, the European Research Consortium for Informatics and Mathematics (ERCIM) in Europe, and the Keio University in Japan.

• The Back Offices: Technical coordinating organizations

Why coordinate? As we've discussed, the Internet is defined by its use of a unified global addressing scheme. The Internet's addressing system consists of two types of identifiers: IP addresses (see [Note 2]), and the domain name system (remember this from the email tale in Part 2?). Each of these identifiers requires global uniqueness, meaning that there can be only one party using 198.65.242.91 on the public Internet, and only party with the domain name "geekcorps.org." It's easy to understand why global uniqueness is important: think of your telephone. If more than one person was assigned your telephone number, your callers would never know whom they will reach when they call your number. Such a system would not inspire confidence. Indeed, zero ambiguity and total reliability are required for the Internet's addressing system if users are going to have confidence that their emails, e-commerce transactions, and secure communications are going to reach their intended destinations.

The need for global uniqueness and reliability in Internet identifiers requires some degree of central coordination. Moreover, there is a finite number of IP addresses -- a matter of particular concern with IPv4, the current version of the Internet protocol -- meaning that they must be allocated and assigned in a coordinated way to ensure that they are not needlessly wasted.

Thus, the Internet relies on a set of technical coordinating organizations, whose common mission is to ensure the availability of globally unique identifiers, and to perform a limited set of critical operational tasks related to those identifiers.

ICANN & IANA. The Internet Corporation for Assigned Names and Numbers (ICANN) is the overall coordinator of the Internet's systems of unique identifiers, including domain names, IP address, and protocol port and parameter numbers, along with the DNS root name server system. Historically, most of these functions were grouped together under the name "Internet Assigned Numbers Authority (IANA)." Since it was founded in 1998, ICANN has been the institutional home of the IANA functions. As such, ICANN is responsible for maintaining and propagating the DNS root zone file. (See Part 2 of our Introduction to the Internet's Infrastructure). In addition to the technical IANA coordination tasks, ICANN serves as the open policy forum for the generic top-level domain name registries (there is no fee to participate in ICANN, any interested person is welcome). Organizationally, ICANN is a small, lightweight organization with a staff of about 20, headquartered in Marina del Rey, California.

Top-level domain registries. The domain name system is highly distributed, meaning that different individuals and organizations administer the top-level domain name registries. Top-level domain registries are generally grouped into two categories: generic and country-code. The country-code registries are designed by two-letter strings, and are associated with countries (like China (.cn), the Netherlands (.nl), Chile (.cl), Mongolia (.mn) or Ghana (.gh)), or geographically distinct territories (like the Faroe Islands (.fo) or Puerto Rico (.pr)). Some of these are run by universities, others by research institutions, governments, non-profits, private companies, or even individuals. The range of geographic and institutional diversity is vast. For a complete list of country-code top-level domains and the organizations that administer them, see the IANA ccTLD database. As the provider of the IANA functions, ICANN is responsible for delegating and re-delegating country-code top-level domains to local administrators, according to the wishes of local Internet communities.

Refer back to Part 2, above, for an explanation of how the domain name system works. For Andrew and Ethan to exchange their emails, can you tell which top-level domain registries were involved?

Regional Internet registries. The regional Internet registries (RIRs) are responsible for allocating and assigning the IP address space. The RIRs are non-profit membership organizations that work to both make IP addresses available to everyone who needs them, while at the same time exercising care to conserve this finite resource. There are currently four RIRs, each covering a defined geographic region:

• Asia-Pacific Network Information Center (APNIC), which covers the Asia-Pacific region, and is headquartered in Brisbane, Australia
• American Registry for Internet Numbers (ARIN), which covers the United States, Canada, parts of the Caribbean, and sub-equatorial Africa, and is headquartered in Herndon, Virginia
• Latin America and Caribbean Network Information Center (LACNIC), which covers Latin America and most of the Caribbean, and is headquartered in Montevideo, Uruguay
• Réseaux IP Européens Network Coordination Centre (RIPE NCC), which covers Europe, the Middle East, Central Asia, and northern Africa, and is headquartered in Amsterdam

In addition, there is a fifth RIR in formation: AfriNIC, which will cover the African continent (currently served by RIPE NCC and ARIN).

The RIRs obtain very large blocks of IP addresses from the IANA, which they then allocate in turn to Internet service providers within their service regions. Those allocations are then sub-allocated until they are assigned to end-user organizations. To varying degrees, the RIRs also provide services for the benefit of the Internet community at large, including databases, routing services, and the coordination of new security and other projects. Each of the RIRs has an open policy forum and various working groups, any of which are free to join by subscribing to the appropriate mailing list, as noted on their websites.

The email exchange between Andrew and Ethan passed through IP addresses assigned by three of the four RIRs. Can you determine which three?

• The Plumbing: Internet Service Providers and Exchange Points

The Internet consists of interconnected networks. The vast bulk of the Internet's capacity is provided by commercial companies that sell Internet connectivity to customers. These companies are generally called Internet service providers (ISPs). The largest ISPs are sometimes called "backbone" ISPs. The service that ISPs sell to their customers is connectivity to every other point on the Internet. Thus, ISPs must cooperate with each other to exchange IP traffic. Facilities at which ISPs exchange traffic are called Internet exchange points.

The email exchange between Ethan and Andrew traversed perhaps a dozen different ISPs and a handful of exchange points. Try performing a traceroute [Note 5] between your computer and some other point on the Internet. Can you determine the ISPs and exchange points that lie between the two machines?

Click here to return to Architecture