FROM THE EDITOR

TERRY R. RAKES, Decision Line Editor Virginia Polytechnic Institute and State University

I would like to begin by welcoming a new feature editor and thanking another who is retiring. Andrew Ruppel, University of Virginia, will be assuming the editorship of the Book Review column. In addition to providing reviews of textbooks, Andy plans to feature non-textbooks which might be of broad interest to our readers. If you have any ideas or suggestions, please pass them along to him. Efraim Turban, who has provided many stimulating articles on Information Resources Management, will be retiring as editor of that column. We thank Efraim for his efforts over the years.

DECISION LINE is now available on the Internet. In this issue, we have included an expanded set of access guidelines (see "Gopher" article). With these instructions, you should be able to access the DSI Internet directory at Georgia State University via your own local gopher, telnet, or telephone access with a modem. I look forward to hearing from you concerning your opinion of the Internet access to DSI information.

In the last issue, we reprinted an article which discussed the availability of Internet access in general at academic institutions, and some of the problems in providing access to those who currently do not have access. In the remainder of this column, we continue our discussion of Internet by presenting an article by Clive Sanford, University of North Texas, who serves as the Institute's computer services coordinator. Clive presents a historical background of Internet and discusses several of its available services.

The Internet: An Electronic Global Village

by Clive C. Sanford, University of North Texas

Connected computer networks are the prototype of a new communications infrastructure that has the potential to be as pervasive as the international telephone network. This technology promises to have effects as far-reaching as those of the postal service, the telephone system, or television. Used extensively by the academic and research communities, the Internet has more recently become commercially available and more accessible to the general public. Dubbed the ``information highway'' by the popular press, the plethora of anachronisms associated with it has led others to call it the ``information hypeway.'' Increased accessibility via ``on-ramps'' or points-of-presence throughout the world has led to more public awareness of the Internet services and its potential for enhancing communications and information retrieval.

The Internet is, therefore, an international meta-network of cooperating, interconnected, multiprotocol networks that support collaboration among thousands of organizations. It has become an integral part of the professional and personal lives of countless people throughout the world. The decentralized structure of the Internet has led to the spawning of some 11,000 separate networks in over 100 countries. Estimates of the Internet's growth are proliferating almost as fast as new host computers and users. It is currently composed of approximately two million host computers and ten million users around the world. While network links are primarily available to countries with a developed infrastructure, other countries in Eastern Europe, South America, and Africa are building "on-ramps" and thus making their resources available to the rest of the world. With the inclusion of many new databases, services, addresses, and projects, the rapid evolution of the Internet does not allow for a concise encapsulation of any one set of commands or maxims. The more you use the Internet, the more you will realize that each day is itself a learning process. The Internet is self-referential; each discovery often leads to another. A casual reference on a mailing list may point to a hitherto unknown resource on a computer halfway around the world. A file on that computer may remind you of a news group which, once subscribed to, updates you on new Internet services. Scattered throughout the world is a broad array of files in a variety of formats and subject matter, ranging from museum exhibits on the Vatican library to photographs of landscapes and people. NASA, for example, has an archive containing imagery from agency missions, including photos of Jupiter, Venus, and the Space Shuttle. There are also collections of shareware and public domain software, and text files on a wide variety of subjects.

The following sections of this article not only address how to use the Internet, but also why the Internet has evolved into a vast interlinked entity made of fiber-optic and copper cable and microwave and satellite links which are capable of connecting the world's fastest supercomputers to 1200 bps dial-up modems.

Internet Protocols

Both the Internet and the telephone system use dedicated telephone lines which are used to establish a connection and transfer information. However, the telephone system uses circuit switching, which employs dedicated channels (more than one channel can be on a physical line) that are only available to the individuals or the computers on both ends of the connection. When a connected channel is not being used it is still unavailable to others until it is released (or disconnected). This leads to an under utilization of network resources.

The Internet is a packet switched network and is modeled after the Postal Service. All correspondence (or packets) is sent together and transferred to the Post Office (or computers) where the packets are sorted and routed to their destinations. On computer networks, this is called packet switching, where the data are divided into packets with headers that contain the necessary routing information. Network software examine these headers and route the data packets to the addressed sites. The computers adjust for failed links or differences in processor speeds by storing the packets in temporary holding areas or buffers. A fundamental advantage of packet switching is that the computers routing the data can select alternate routes if a link fails. Also, computers at either end of a packet network can operate with different protocols.

A protocol is a set of rules that determines how data will be exchanged between different computers. Their use allows for the creation of standards which are not tied to a particular hardware system. One of their most important functions is to specify how the network is to move messages and handle errors between resources. Protocols establish naming conventions of a host, mailbox, or other resource identifiers. The address specifies the logical location of a resource and supplies routing information for the network software. Internet addresses have four numbers, each less than 256, that are separated by periods. For example, an Internet host named jove.acs.unt.edu has an Internet address of 129.120.1.41.

The IP (Internet Protocol) hierarchically handles the naming, addressing, and routing of packets. For example, jove.acs.unt.edu is an Internet Domain Name Scheme (DNS), where EDU is a top level domain, unt.edu is a subdomain of EDU, and acs.unt.edu and jove.acs.unt.edu are further subdomains (in this case, jove.acs.unt.edu is a host machine). The IP address 129.120.1.41 is a two-level address, where the prefix 129 is a network number and the rest (the local part) is a host number. Different networks apply their own mapping techniques of the local part to network addresses.

Unfortunately, when a resource is connected to more than one network, it may have more than one name, address, or route. For instance, unt.edu may be known as untvax on the UUCP (UNIX-to-UNIX Copy Program) network.

The IP limits information within a packet to be from 1 to about 1500 characters long. Since, in many cases, the information that needs to be transmitted is longer than this, the TCP (Transmission Control Protocol) breaks the information into packets. It then numbers each packet so that receipt can be verified and the data can be reassembled back into proper order at its destination. It therefore provides reliable, ordered, end to end delivery of IP packets.

TCP/IP, or Transmission Control Protocol/Internet Protocol, is therefore a system of protocols whose success has made the Internet possible. These protocols were developed in 1974 by Robert Kahn, a major figure during the ARPANET development, and is now president of the Corporation for National Research Initiatives (CNRI), and computer scientist Vinton G. Cert, now president of the Internet Society and vice-president of CNRI. Systems using other protocols, such as BITNET, connect through gateways (specialized computers that user protocol converters) into the Internet, while the UUCP network, which connects thousands of computers by dial-up telephone lines, makes its electronic mail destinations available to Internet users.

This inherent flexibility and decentralized structure was made possible about 25 years ago when the Advanced Research Projects Agency (ARPA) of the U.S. Department of Defense funded one of the first computer networks.

Historical Origins

In 1969, the birth of the Internet occurred when the ARPANET (Advanced Research Projects Agency Network) was designed and developed by Bolt, Beranek, and Newman under contract with ARPA. The ARPANET was constructed so researchers could share information with university, military, and defense contractors and to study how communications could be maintained in the event of partial outages which could occur during a nuclear attack. Since other networks were also being assembled at this time, it became clear that new communication methodologies were necessary. In 1973, ARPA, under its new acronym DARPA (Defense Advanced Research Projects Agency), initiated the Internetting Project in order to study how to link packet networks together. The central theme of the design was to minimize the information required from the computers that were accessing the network. Data were organized in Internet Protocol (IP) packets, or envelopes with destination addresses. The use of these addresses allowed for the responsibility of error free transmission to be shifted from the communication links to the computers. Peer-to-peer communications between the network computers allowed for information exchange between computers to occur on an þequalþ basis. The IP software became central to the concept of internetting, or networking different networks together. It answered the need to allow network-specific protocols to communicate via gateways which connect networks with a heterogenous mix of hardware and software. Sites with LANs (Local Area Networks) could now connect directly to the ARPANET instead of being constrained to connecting to a single large timesharing computer.

When DARPA adopted the TCP/IP protocol for all ARPANET hosts (computers), it established a standard for the Internet. This introduced flexibility and allowed for expansion of the network through the addition of gateways to other networks. In 1980, CSNET (Computer Science Network), a network linking computer science departments in several states, was the first autonomous network DARPA allowed to connect to the ARPANET. CSNET eventually merged with BITNET (Because Its Time Network) in 1989. In 1983, MILNET (Military Network) was split from the ARPANET. The ARPANET was used for continuing research on networking until June of 1990, when it was decommissioned. By then, the main principle of internetworking had been established, and its functions were absorbed into the broader structure of the Internet. Networks were communicating with a set of workable protocols, and new networks being added into an expanding metanetwork interconnected through gateways. By 1990, the National Science Foundation had taken over much of its functionality through a network of its own.

The National Science Foundation Network, or NSFNET, grew out of a need to connect six supercomputer centers around the country. It adopted the ARPANET's IP technology. In 1986, the NSFNET backbone (primary network) connected the six sites by 56-kilobits-per-second (Kbps) lines, which quickly became overloaded as traffic increased. By July, 1988, the network had 13 nodes (computers) using T-1 connections (leased lines) operating at 1.5 megabits per second (Mbps). In September of 1990, the NSF formed Advanced Networking Services, Inc. (ANS), a coalition of the Merit (the Michigan Education and Research Infrastructure Triad), IBM, and MCI companies. Operating under contract with Merit, the ANS managed the T-1 backbone for NSFNET and built a new T-3 (45 Mbps) backbone to replace it. This backbone was put in operation in December of 1992, thus creating a 700-fold increase in throughput. T-3 speeds carry data at the equivalent of fourteen hundred pages of single-spaced typed text per second. NSFNET data traffic had grown from 195 million packets per month during August 1988, to almost 24 billion per month by November 1992.

With the T-3 backbone functioning, a new arrangement was developed that allowed ANS to operate two separate networks over the same equipment. NSFNET continued to support institutions reliant on government subsidies, and a subsidiary dubbed ANS CO+RE supported commercial users of the network. In doing this, ANS joined UUNET (Unix-to-Unix Network) Technologies and Performance Systems International (PSI) in the ranks of commercial providers.

As another ANS customer, NSFNET has become a national research network involved in the creation of a National Research and Education Network (NREN). It also links agency networks such as ESnet (Department of Energy) and NSInet (National Aeronautics and Space Administration), as well as providing access to local and regional networks. Some mid-level networks, such as THEnet (Texas Higher Education Network) are operated by their states, and others are run by university consortia operating large regional networks, such as SURAnet (Southeastern Universities Research Network) or MIDnet (Midwest TCP/IP Network). Some of the regionals are run commercially, as is CERFnet (California Education and Research Network), while others are managed by university computer scientists.

Each network pays for their part of the Internet. For example, the NSF pays for NSFNET and NASA pays for NSInet. Network administrators collaborate on how to interconnect and fund the interconnections. Some networks have become independent from NSF subsidies while others retain close NSF connections. NSF has encouraged regional networks to connect to new sites with the general expectation that the regionals will become financially independent within three to five years. The regionals have also been encouraged to offer network services in order to help them achieve independence. The Internet is now an internetwork of many networks all running the TCP/IP protocol suite connected through gateways and sharing common name and address spaces. By this definition, other networks which use different protocols, such as BITNET and UUCP, are therefore not part of the Internet. In many cases, though, references to the Internet imply access to the Internet and other networks that run under different protocols.

Internet Services

Application programs which implement the TCP/IP protocols provide end-users with network services. Internet services fall into one of four categories: (1) computer mediated communication (CMC) which use SMTP (Simple Mail Transfer Protocol), (2) file transfer with FTP (File Transfer Protocol); (3) remote login with interactive sessions (Telnet); and (4) tools which implement other protocols and client/server models. Many of the network services expedite and coordinate many tasks in the scientific community and attract people who are involved in similar activities. These services are designed to attenuate search costs and enhance the interaction between individuals and groups.

Electronic Mail

The most widely used service involves computer mediated communication (CMC). The primary effect of CMC is increased human interaction, which can lead to better technical productivity through the exchange of ideas and references. CMC services allow people to exchange messages, and may be either one-to-one (mail), one-to-many (distribution lists or bulletin boards), or many-to-many (news or true conferencing systems). These services can be batch (asynchronous) or interactive. A message may be delivered and read immediately in an interactive service or after a delay in a batch service. Batch CMC does not require immediate action on the part of participating users or supporting programs and protocols. Thus, asynchronous communication requires neither dedicated connections between machines nor simultaneous interaction between individuals. The most obvious advantage of batch services is that the recipients of a message do not have to be actively participating when a message is sent and hence circumvents the annoyance of ``telephone tag.'' For detailed discussions which do not require an immediate response, mail or batch conferencing may be the best service to use. A true conferencing system is most applicable when the number of participants, schedule differences, or the urgency of the situation increases.

File Transfer

FTP (File Transfer Protocol) is a capability of almost all computers which have access to the Internet. The FTP command or program allows you to establish an interactive file transfer session with a remote computer system on the Internet. FTP usually allows you to view a directory of files on the remote system, change directories, get a file from a remote system, or in some cases, put a file from your local system to the remote system's directory.

Many Internet hosts implement a service called Anonymous FTP which allows anyone on the Internet to establish a file transfer session by using the user name anonymous. In most cases the files available are publicly-distributed documents, shareware, or freely-distributed software. The actual command to use FTP varies among operating systems, but 'ftp' is usually the most common command. For example:

Telnet

A number of Internet services can be accessed by Telnetting to an Internet host. Telnet is the program or command which allows you to establish an interactive session with a remote system. It allows you to log in to a remote system as if it were a local host. The actual command for negotiating a Telnet session varies between systems, but the most common is the word 'telnet'. For example:

Tools

Resource discovery is a major theme of the network, and there is a broad range of tools for this. The use of a developing generation of resource discovery tools will, for example, allow for searching an Internet database with new techniques such as Wide Area Information Services (WAIS). However, we have not yet reached the point where different approaches to presenting information have all been subsumed under a single user interface.

A number of universities are making various information services accessible via the Internet. The three most common types of services are on-line library card catalog systems, bulletin board systems (BBSs), and campus-wide information systems. Like on-line catalog systems, Campus-Wide Information Systems (CWISs) are available through a Telnet connection. CWISs usually use menu-driven software to organize information relevant to a particular school and are made available to other interested parties via the Internet. A number of services don't require you to explicitly establish an FTP or Telnet session. These applications will often either FTP or Telnet from a menu system and retrieve and/or organize information. They sometimes use a client/server model in which some processing of data occurs on both the local and remote hosts. A client is a software application designed to interact with a service somewhere else on the network (the server).

Services that employ a client/server architecture include archie, gopher, and WAIS. Archie is an archive server, or a database which manages names of files and the sites on the Internet where they are available for anonymous FTP. Usually, archie clients can be invoked by typing archie at your system prompt. If you do not have an archie client program available at your site, you can Telnet to archie.sura.net, log in as archie, and follow the instructions to query the archie database. Gopher, better known as the furry rodent that is the mascot of the University of Minnesota, makes it easy to access Internet resources such as files, telnet sites, or WAIS (Wide Area Information Services) databases. WAIS þclientþ programs allow you to enter search parameters which query WAIS server databases and, when a match is found, returns information on the topic you specified. Usually, gopher clients can be invoked by typing gopher at your system prompt. If you do not have a gopher client program, you can telnet the consultant.micro.umn.edu server at the University of Minnesota and log in as gopher. It is anticipated that the number of client/server applications available on the Internet will see dramatic growth over the next several years, especially as large-scale collections of on-line information becomes more prevalent and as the network grows to accommodate additional use.

A wide range of skills must be employed to use the individual network resources. Therefore, when using the Internet, it is important to decide on your objective and then select your method of access. If you want to log in to another system on the Internet, use Telnet; if you want to transfer a file, use FTP; if you are looking for a file or program on the Internet, try archie; if you are wanting information about a particular university, check to see if they run a CWIS or BBS; if you need on-line information on a topic, try gopher.