Research

Networkings

Ask Gerry Santoro how the Internet works and heíll reel off a string of analogies. First of all, he stresses, it works like any other system for communicating between human beings. "We tend to be amazed by the flash of technology, but the whole purpose is in the information communicated to a user by it—or by a user to it. That information flows between people."

Santoro, who is both lead research programmer for Penn Stateís Center for Academic Computing and an affiliate assistant professor of speech communication, gave the fourth of this yearís Penn State Lectures on the Frontiers of Science.

global computer network map

Historically, he noted, the global computer network now known as the Internet grew out of Defense Department experiments conducted in the late 1960s, when researchers at the RAND Corporation were asked to build a communications system that could survive a nuclear war. "The challenge was to create a network where all the nodes were independent," he explained, "so that even if a large part of the system were destroyed, messages would still reach their destinations."

Packets of Info

The solution was a concept called "packet switching," in which streams of digital data are broken into small "packets," each containing about 200 "bytes" of information. (A byte is a set of eight bits; a bit is a single digit, a ë1í or a ë0.í) Each packet is addressed and sent through the system on its own, traveling node to node, forwarded by successive computers along the best route depending on traffic conditions.

"Itís like mailing a series of postcards, each containing part of a message," Santoro said. At the end of their separate journeys, the cards are reassembled into a coherent whole.

The beauty of packet-switching, Santoro noted, is that, "It doesnít matter how the cards move through the system. They can arrive successfully by many different routes. "If one post office happens to be closed, the cards can be rerouted through another." Packet-switching is cheap, too, since communicating doesnít require an exclusive, closed channel between two parties. Packets can be placed into the system by many users at once, and are simply ferried along in the order received.

The first generation of this new concept was ARPANET, a network joining employees of the Defense Departmentís Advanced Research Projects Agency. By the early í70s, ARPANET offered these cyberpioneers three nifty tools that today are taken for granted. By following a set of programming rules dubbed Simple Mail Transfer Protocol, or SMTP, they could send each other mail electronically. A second set of rules, File Transfer Protocol (FTP), allowed them to exchange files of data. Finally, a program known as Telnet permitted them, with passwords, to log on to their own computers from any other computer on the system.

Following Protocol

Before long, Santoro said, the Army, Navy, and Air Force all had their own networks—each developed by a different computer company. "The question became, How do you interconnect all these different types of equipment?" The answer lay in developing yet another set of standards, an overarching protocol "suite" called TCP/IP. TCP, for Transmission Control Protocol, would standardize the conversion into packets, and their reassembly into messages. IP, for Internet Protocol, would provide an identifying number for every machine on the system and route packets where they needed to go. Together, TCP/IP would function as a sort of common language, laying the groundwork for a worldwide system. Strictly defined, Santoro said, todayís Internet is "the interconnected set of all networks running TCP/IP, supported by packet switching."

From a technical perspective, he went on, "the Internet works exactly like a telephone network. A phone system requires only two things," he explained. "Every telephone has to be electrically connected to every other telephone, and each telephone has to have a unique identification number. The Internet follows the same strategy—in fact, a lot of times they share the same wires. And every computer has its own identifier—the IP number. When you run a program like a Web browser, and you select a link, whatís happening is your computer is going out and making contact with some other computer, and the two computers are exchanging packets of information."

Say you want to see the Web version of Santoroís talk, for instance. Within your browser, you type in the file locator he assigned it: http://cac.psu.edu/~santoro/internet/. When you hit "Enter," the first thing your browser does is to seize on the part of that locator called the domain name, which corresponds to the IP number of the computer on which the lecture resides. ("Human beings remember names better than numbers," Santoro explained.)

Your browser passes the domain name, cac.psu.edu, to the TCP/IP software on your computer. Your TCP/IP software contacts a domain-name server, a computer that acts like a phone book, containing a long list of domain names and corresponding IPs. Once your computer gets the correct number, it places the "call."

When the remote computer—in this case a machine in Penn Stateís Academic Computing building—answers, your browser sends it a request for the specific file: ~santoro/internet/. Through its own TCP/IP software, the remote computer then begins sending the necessary packets, and when all have been received and reassembled, your web browser displays the file on your screen. Presto! Thereís the welcome page, complete with photo image of a smiling Santoro in flannel shirt and shades.

Hooking Up

Your actual physical connection to the Internet can be any of several types: a modem connected to your home telephone line, a fiberoptic cable linking several computers in your school or office, even a satellite hookup. For now, Santoro said, the most popular type remains the modem, a device which converts outgoing digital data into an analog signal—that shrill duotone you hear when you mistakenly pick up the telephone receiver in midtransmission. The analog signal travels the telephone line to another modem at the computer of your Internet service provider, where it is reconverted to packets of 1s and 0s. (Switching from digital to analog is called modulation: "modem" is short for modulator/demodulator.)

Your local provider then merges your packets with those arriving from other users (a step known as "multiplexing") and sends them along to a larger, centralized provider, from where they can be shoved out onto a "backbone"—a high-speed transmission line—to zip across the country, or the world. When they reach a local provider near their intended destination, the initial process is reversed. Finally your packets arrive, all in a heap, at the server—that distant computer serving up the text file or video clip you desire.

On the Road

A trip round the world in cyberspace takes but a couple of blinks. When things go right, that is—hereís where Santoro pulled out the most well-worn Internet analogy of them all. "From a behavioral perspective," he said, "the Internet does work very much like a highway system. It has its rush hours, its traffic jams, its weather and construction delays. To be an effective user, youíve got to take these realities into account.

"No one anticipated the traffic levels we have today," Santoro said. And traffic doubles every 100 days, he added, as more and more people flock to the World Wide Web.

A group of academic, industry, and government partners, including Penn State, is hard at work developing the infrastructure for Internet 2, a new, superfast network that will relieve some of this space crunch, Santoro noted. But he doesnít see band-width as the biggest restricter of the Internetís huge potential.

"The real limiting factor, I think, will be the ability of individual users to deal with all the information thatís going to be coming their way," he concluded. "Learning how to make sense out of it."

Gerald M. Santoro, Ph.D., is lead research programmer in the Center for Academic Computing and affiliate assistant professor of speech communication in the College of the Liberal Arts, 215C Computer Building, University Park, PA 16802; 814-863-7896; gmsantoro@psu.edu.

Last Updated May 1, 1999