Columbia University
Information Technology

Networks | Details | Domain Name Server | FTP

Telnet | E-mail


A hardware protocol used to connect an Internet host to a packet switch on the ARPANET and MILNET. This protocol is also called AHIP (Asynchronous Host Interface Protocol). The number 1822 comes from the BBN (Bolt Beranek and Newman) report that defined the interface for the original ARPANET. 


Short for acknowledgement. 


A type of message sent to indicate that a block of data arrived at its destination without error. A negative acknowledgement (NACK) indicates that the block of data was not correctly received. 


American National Standards Institute. This organization is responsible for approving U.S. standards in many areas, including computers and communications. Standards approved by this organization are often called ANSI standards (e.g., ANSI C is the version of the C language approved by ANSI). ANSI is a member of the International Standards Organization (ISO). 


A networking protocol developed by Apple Computer for communication between Apple Computer products and other computers. This protocol is independent of what network it is layered on. Current implementations exist on Localtalk (a 235-kilobit/second local area network (LAN)), and Ethertalk (a 10-megabit/second local area network). 


Address Resolution Protocol. This protocol is used to dynamically bind an Internet address to a low-level physical network address. It is often used on local area networks (LANs) such as Ethernet. 


One of the first heterogeneous-host packet switching networks developed for the Advanced Research Projects Agency of the Department of Defense (see DARPA). The ARPANET became operational in 1968; it was the proving ground for many of the protocols and concepts in todayUs Internet. 

Authority Zone

The part of a domain name that a single name server resolves. For example, if the server spooler is responsible for resolving all machine addresses in the domain, then its authority zone is * (where * means anything is allowed). On the other hand, would not be in its authority zone. 

Autonomous Confederation

A group of independent computer systems that trust each other regarding routing (see route) and reachability information. Members of an autonomous confederation will believe information provided by other members of the confederation in preference to information received from systems that are not part of the confederation. 

Autonomous System

A collection of networks controlled by one administrative authority. The gateways within this system are expected to trust one another and to share and update routing information (see route) among themselves by any mutually agreeable protocol. A core gateway must also be designated to share routing information with other autonomous systems via EGP. 


A central high-speed network connecting independent subnetworks. Today, the NSFNET provides a backbone network for regional networks such as NEARnet, CSNET, and BARRNet. 


The frequency width of a communications channel, usually measured in hertz, kilohertz, or megahertz. For example, one channel on a satellite transponder might have a bandwidth of six megahertz, thereby enabling it to carry a television signal. Sometimes, this term is applied to how much digital information a channel can carry, usually in conjunction with fully digital communications lines. For example, a T1 line might be said to have a bandwidth of 1.544 megabits/second; however, it would be more correct to say that a T1 line can carry or transmit 1.544 megabits/second. 


A transmission medium where digital signals are sent without complicated frequency shifting. In general, only one communication channel is provided at a time on a baseband system. Ethernet is a baseband network. 


The number of symbols that may be sent over a communications channel per second. Each symbol may be an arbitrary analog signal, and it may represent more than one bit of information. For example, a communications channel transmitting at 2400 baud, with each symbol containing four bits, is capable of sending 9600 bits per second (this is in fact the way V.32 9600-baud modems work). 


Bolt Beranek and Newman Inc., a diversified high-technology company in Cambridge, Massachusetts, was awarded the original contract to build the ARPANET and has been extensively involved in Internet development. Today, BBN is responsible for managing the NNSC, CSNET, and NEARnet among others. This stack is brought to you by the NNSC staff at BBN (Hi Mom!). 


Because ItUs Time Network. An academic and research network connecting approximately 2500 computers in thirty-two countries. This network provides interactive electronic mail, and file transfer services via a store-and-forward methodology based on IBM NJE protocols. BITNET traffic and Internet traffic are exchanged via several gateway hosts. This network is now part of the Corporation for Research and Educational Networking (CREN). 


A transmission medium where multiple digital channels are frequency multiplexed onto a single cable. This type of network requires relatively complicated electronics, but is capable of carrying voice, data, and video all on the same medium. Cable television systems are examples of broadband networks. 


A technique used to send packets to all hosts on a network. Broadcasts are often used in conjunction with ARP and RARP protocols on local area networks. 


Berkeley Source Distribution. This acronym is used to describe the versions of the UNIX operating system and its utilities developed and distributed by the University of California at Berkeley. "BSD" is usually preceded by the version number of the distribution, e.g., "4.3 BSD" is version 4.3 of the Berkeley UNIX distribution. Many Internet hosts run BSD software, and it has been the ancestor of many commercial UNIX implementations such Sun OS and SequentUs Dynix. 


A term coined to describe the communications structure created when packet switched networks are connected by gateways. The term internet without a capital I is now more commonly used. 


Comit Consultatif International de T l graphique et T l phonique (International Consultative Committee on Telephone and Telegraph). This organization is part of the United Nations International Telecommunications Union (ITU) and is responsible for making technical recommendations about telephone and data communication systems. X.25 is an example of a CCITT recommendation. Every four years CCITT holds plenary sessions where they adopt new standards; a session is planned for 1992. 


A computed symbol whose value is dependent upon the entire contents of a message or packet. This value is usually sent along with the message when it is transmitted. The receiving system computes a new checksum based upon the received data and compares this value with the one sent with the packet. If the two values are the same, the receiver has a high degree of confidence that the data was received correctly. 


A computer system or process that requests a service of another computer system or process. A workstation requesting the contents of a file from a file server is a client of the file server. 


An agreement between two processes or hosts to pass information along a specified protocol path without further exchanges of addressing information. 


Corporation for Open Systems. An international non-profit organization made up of computer users and vendors. This organizationUs mission is to provide ways of testing OSI implementations. 


Corporation for Research and Educational Networking. This organization was formed in October, 1989, when BITNET and CSNET were combined under one administrative authority. CREN is now responsible for providing networking service to both BITNET and CSNET users. 


Carrier Sense Multiple Access with Collision Detection (phew!). This is a characteristic of a local area network (LAN). When multiple users have access to the network for transmitting data, the network avoids transmitting data from more than one user at a time, so that they avoid running into each other. Ethernet works this way. 


Computers and Science Network. A network that was established to provide mail forwarding and Internet connectivity to computer (and now other) science researchers. This network primarily provides electronic mail service via dial-up lines, although X.25 and Internet services are available from sites that are suitably connected. This network is now part of the Corporation for Research and Educational Networking (CREN). 


Defense Advanced Research Projects Agency. An agency of the U.S. Department of Defense responsible for the development of new technology for use by the military. DARPA (formerly known as ARPA) was responsible for funding much of the development of the Internet we know today. The New York Times business section called DARPA "AmericaUs answer to JapanUs MITI." 


A packet whose routing (see route) and interpretation is independent of other packets being sent by that host. Every datagram must contain a destination address, since it cannot rely on addressing information sent by previous packets. Datagrams are a connectionless form of communication, and are the basic building blocks of the internet protocol (IPQsee TCP/IP). 


Defense Data Network. A worldwide operational communications network serving the US Department of Defense composed of ARPANET, MILNET, and other portions of the Internet, used to connect military installations. It is run by the Defense Communications Agency (DCA). 


European Academic Research Network. A network connecting European university and research institutions providing electronic mail and remote job entry facilities. This network uses BITNET protocols and connects to BITNET in the U.S. 


Exterior Gateway Protocol. This protocol is used by a gateway representing an autonomous system to export to other gateways information concerning networks and gateways contained within that system. 

Electronic Mail

A system whereby a computer user can exchange messages with other computer users (or groups of users) via a communications network. Electronic mail is one of the most popular uses of the Internet. 


A 10-megabit/second standard for local area networks (LANs), initially developed by Xerox, and later refined by Xerox, DEC, and Intel. All hosts are connected to a coaxial cable where they contend for network access according to the CSMA/CD protocol. 

Email (or E-mail)

Shortspeak for electronic mail (q.v.). 


Fiber Distribution Data Interface. A newly emerging standard for a fiber-optic local area network (LAN) running at 100 megabits/second. 


In computer messages, data files, and programs, a field is a group of characters that is treated as a unit. For example, each TCP/IP packet contains fields for addressing and routing information (see route). 

Internet users may encounter fields in the header of an electronic mail message. The fields are lines that begin with a field-name followed by a colon and a space. To: and From: are the only required header fields, but there are optional standard fields for the user, and fields that are added by the mail delivery system. The format of email messages is defined in RFC-822. 

File Server

A computer whose principal purpose is to store files and provide network access to those files. 


A piece of a packet. When a gateway is forwarding a maximum size IP (see TCP/IP) packet to a network that has a smaller maximum packet size, it is forced to break up that packet into multiple fragments for transport on the new network. These fragments will be reassembled by the IP layer at the destination host (or possibly by an intermediate gateway under some circumstances). 


An assembly of bits at the Data Link layer of the ISO protocol stack. This collection of bits begins with some bits used for header information, and ends with some checksum bits used for error detection and/or correction. All bits between the header and the checksum are data. 


File Transfer, Access, and Management. An application layer protocol for moving and manipulating files. 


File Transfer Protocol. A protocol permitting a user on one Internet host to access and transfer files to another host over a network, such as the Internet. FTP is usually the name not only of the protocol, but also of the program the user invokes to execute the protocol (e.g., ftp This protocol is usually layered on top of TCP and IP (see TCP/IP). FTP is available on several operating systems. You can use the ftp command to copy computer files that contain a variety of information, such as software, documentation, or maps. 


A computer used to connect together one or more networks. This computer is seen as a host by the networks to which it is connected, but is capable of forwarding packets from one network to another. Gateways are also responsible for providing and receiving routing information to other gateways in the Internet so that they will know the best routes for sending packets between networks. One may think of a gateway as a packet switch with whole computer networks as its communication links. 


Government OSI Profile. GOSIP is a collection of ISO specifications for mixed-vendor networks for use by the government. Government networks are mandated to support GOSIP in the not-too-distant future. 


The header is information that appears at the top of an electronic mail message. See field. 


An inter-packet switch protocol used in the NSFNET to determine shortest delay routing (see route). This protocol is only used among packet switches that trust each other. 


A computer that allows users to communicate with other host computers on a network. Individual users communicate by using application programs, such as electronic mail, TELNET, and FTP. 


Internet Control Message Protocol. This protocol is an integral part of the Internet Protocol (IPQsee TCP/IP). The protocol is used to exchange error and control information among IP hosts. For example, a gateway that is sent an IP datagram for which it is not the best route would send an ICMP redirect packet back to the originating host to inform it of the best route. ICMP implementations also provide fault isolation capabilities such as packet echo. 

IEEE 802

The IEEE standards for local and metropolitan area networks (see LAN and MAN). This class of standards is further broken down by type of network, each of which is specified by digits after a decimal point. For example, the Ethernet standard is 802.3; IBM Token Ring is IEEE 802.5. 


This stands for Internet Engineering Notes. 


Interface Message Processor. This was the name for the original packet switches used in the ARPANET and MILNET. Today, the term Packet Switch Node or PSN is in more common usage. 

Internet Address

A thirty-two-bit number that uniquely identifies an Internet host. This address is typically represented in eight-bit numbers (octets) separated by dots, e.g., An Internet address consists of a network number and a host number, and may be a class A, B, or C address. A class A network address is formatted as N.H.H.H, providing seven bits of network number and twenty-four bits of host number (e.g., indicates host 117 on net 26). A Class B network address is formatted as N.N.H.H, providing fourteen bits of network number and sixteen bits of host number (e.g., indicates host 1.132 on net number 128.89). A Class C network address is formatted as N.N.N.H, providing twenty-two bits of network number and eight bits of host address (e.g., indicates host 28 on network number 192.1.14). 

The Internet is the interconnection of many networks throughout the world that speak the same language, namely the TCP/IP protocol suite. Internet with a capital I refers specifically to that internet that contains NSFNET, MILNET, and DDN. 

You may see "internet" with a small "i." This can refer to any network built out of the TCP/IP protocol suite, or it might refer to networks using other protocol families that are composites of smaller networks. 


Integrated Services Digital Network. A public digital network designed to integrate voice and non-voice traffic. This system is intended to be a replacement for our current analog telephone systems, and as such is being standardized by the CCITT. 


International Standards Organization. The international body responsible for establishing multivendor networking standards. 


Communication networks for computers may be organized as a set of more or less independent protocols, each in a different layer (also called level). The lowest layer governs direct host-to-host communication between the hardware at different hosts; the highest consists of user applications. Each layer builds on the layer beneath it. For each layer, programs at different hosts use protocols appropriate to the layer to communicate with each other. 

TCP/IP has five layers of protocols, and OSI has seven. The advantage of different layers of protocols is that the methods of passing information from one layer to another is specified clearly as part of the protocol suite, and changes within a protocol layer are prevented from affecting the other layers. This greatly simplifies the task of designing and maintaining communication programs. 


Local Area Network. A data network intended to serve an area of only a few square kilometers or less. Because the network is known to cover only a small area, optimizations can be made in the network signal protocols that permit data rates in the 10-megabyte-per-second to 100-megabytes-per-second range today. Wide-area communication is accomplished by connecting LANs together via metropolitan area networks (MANs) or wide-area networks (WANs). Both Ethernet and FDDI are local area networks. 


A local area network (LAN) protocol developed by Apple Computer. This network is designed to run over twisted pairs of telephone wire and has a data rate of 235 kilobits/second. All Macintosh computers contain a LocalTalk interface. 

Mail Bridge

A mail gateway that forwards electronic mail between two or more networks while ensuring that the messages it forwards meet certain administrative criteria. A mail bridge is simply a specialized form of mail gateway that enforces an administrative policy with regard to what mail it forwards. 

Mail Gateway

A network host that forwards electronic mail between two or more possibly dissimilar networks. In the process of forwarding the mail, the gateway may have to reformat addresses and mail headers to conform with the electronic mail standards of the destination network. 


Metropolitan Area Network. A data network intended to serve an area approximating that of a large city. Such networks are being implemented by innovative techniques such as running fiber cables through subway tunnels. 


Manufacturing Automation Protocol. A protocol stack developed by General Motors following the OSI model that guarantees access to each host within a certain maximum time. At the upper layers, it includes many of the OSI standards. At the lower layers, it is based upon Token Bus (IEEE 802.4). 


"Message" has multiple meanings: 

  • A user-defined collection of data sent over a network. 
  • A piece of text displayed on a terminal screen that was sent by a user or a program. 
  • A collection of data sent from one computer programming entity to another.  

MILitary NETwork. This network was created in 1984 from parts of the original ARPANET. The military users wished to have an operational production network, while the research community wished to have a network on which to continue experimenting in networking. Therefore, the military users were placed on MILNET, the research users were placed on ARPANET, and the two networks were connected with mail bridges and gateways. Today, MILNET is one of the class A networks in the Internet. 


Maximum Transmission Unit. The largest number of bits that a network permits to be transmitted as one packet. 


National Bureau of Standards. This organization, which was part of the U.S. Department of Commerce, was responsible for establishing standards in the United States. It has since become the NIST. 


A computer network is a group of computers that can communicate electronically. Networks can be composed of computers in a single building (Local Area Networks or LANs), or computers thousands of miles apart (Wide Area Networks or WANs). The Internet is a worldwide collection of computer networks that can intercommunicate. The system manager and computer center staff at your site can provide information about your local network. 

Network Address

A number or group of numbers that uniquely specifies a host on a network. For example, is the network address for Also, informally, an electronic mail address. For example, is the network address for the NSF Network Service Center (NNSC). 


Network File System. This acronym describes a protocol developed by Sun Microsystems to allow a computer system to access files over a network as if they were on its local disks. This protocol has been incorporated in products by more than two hundred companies, and is now a de facto Internet standard. 


This stands for the National Institute for Standards and Technology (see NBS). 


Network Operations Center. A location from which the operation of a network or internet is monitored. This center also usually serves as a clearinghouse for problems and efforts to resolve those problems. 


National Science Foundation. A government agency whose purpose is to promote the advancement of science. NSF funds science researchers, scientific projects, and infrastructure to improve the quality of scientific research. The NSFNET, funded by NSF, is an essential part of academic and research communications. 


Network Time Protocol. A protocol built on top of TCP (see TCP/IP) that assures accurate local time-keeping with reference to radio and atomic clocks located on the Internet. This protocol is capable of synchronizing distributed clocks within milliseconds over long time periods. 


Office Document Architecture. This emerging standard defines ways in which text, graphics, and facsimile documents can be moved over a multivendor network. 


Open Systems Interconnect. Usually used as shorthand for the Open Systems Interconnection Reference Model (OSI Reference Model). 

OSI Reference Model

A seven-layer structure designed to describe computer network architectures and the way that data passes through them. This model was developed by the ISO in 1978 to clearly define the interfaces in multivendor networks, and to provide users of those networks with conceptual guidelines in the construction of such networks. 


A collection of data sent as a unit along a packet network. Packets are self-contained; each packet has its own source address and destination address and cannot exceed a maximum size. Long messages are broken up into multiple packets for transmission over the network. 

Packet Switch

See PSN. 


Packet Assembler/Disassembler. A network host designed to interface terminals to a packet network. 


Packet Internet Groper. A program that sends packets to a remote host on the Internet and looks for replies. This program works via the echoing facility provided by the ICMP protocol and is a way to determine if an Internet host is reachable from your host. 


A mutually agreed procedure for communicating information between two parties. Standard protocols are the basis for all computer communication. 


Packet Switch Node. A dedicated computer whose purpose is to accept, route, and forward packets in a packet switched network. 


Reliable Datagram Protocol. An Internet standard protocol for reliably sending datagrams between user programs. This protocol is like UDP, but guarantees delivery and does retransmission as necessary. This protocol is built on top of IP (see TCP/IP) and uses IP for datagram delivery. 


An obsolete version of the Request for Comments (Standard for the format of ARPA Internet Test Messages, August 16, 1982) that specifies the format of electronic mail messages. See RFC-822. 


The current version of the Request for Comments that specifies the format of electronic mail messages. 


Request for Comments. RFCs are the principal documents used on the Internet to propose new protocols and services. These documents are published as electronic documents on by the DDN NIC. 


Routing Information Protocol. A routing (see route) protocol provided in the Berkeley UNIX (see BSD) operating system, that permits a group of hosts located on a local network to share routing information. This function is provided by the program routed. 


A path from one Internet host to another. 


Remote copy. A program and protocol provided in the Berkeley UNIX operating system (see BSD) that permits files to be copied from one computer to another by an extension to the syntax of the UNIX cp (copy) command. This protocol is largely implemented among UNIX machines, but the protocol is general enough that non-UNIX machines may use it. However, rcp does not provide the word-length adaptability and flexibility that the FTP protocol does. 


Remote login. A program and protocol provided in Berkeley UNIX (see BSD) that permits a user on one computer to log in to another computer. This protocol is largely implemented among UNIX machines, but the protocol is general enough that non-UNIX machines may use it. For example, Excelan ANNEX terminal concentrators permit users on dumb terminals to use the rlogin protocol to communicate with Internet computers. 


A device that chooses routings for packets. This is a generic term and applies to such diverse devices as bridges (which pass packets from one physical LAN to another with almost no interpretation) and WAN gateways (which pass packets from one wide area network to another, doing fragmentation and reassembly as necessary). 


A computer system or process that provides a service for other computer systems or processes to access. A supercomputer can be thought of as a computation server. A program that provides Internet File Transfer Protocol (FTP) access to local files is usually called an FTP server. 


Simple Mail Transfer Protocol. This Internet standard network protocol is used to move electronic mail messages from one host to another. 


Systems Network Architecture. A proprietary networking architecture used by IBM and IBM-compatible mainframe computers. Because of its widespread use, SNA is a de facto standard. While it can use packet switched networks for transport, SNA is largely a circuit-switching rather than a packet-switching technology. 


Simple Network Monitoring Protocol. This Internet standard protocol is used by a network monitoring center to gather information regarding the status of hosts on its network or on the Internet. 

Source Address

The network address of the host that originates a packet. 


Standards Promotion and Applications Group. This European organization collaborates with COS to promote testing procedures and techniques for OSI products. 


A computer responsible for routing (see route) packets in a packet switched network.


A communications service over leased lines and microwave links that runs at 1.544 megabytes per second. The major links of the NSFNET are T1. Faster services such as T3 (45 megabytes per second) are available, although they are not yet off-the-shelf products. The NSFNET is in the process of upgrading to T3, and plans much higher transmission rates for the future. 


Transmission Control Protocol/Internet Protocol. A Department of Defense standard protocol suite encompassing both network and transport level protocols. While the terms TCP and IP specify two protocols, common usage of the two terms together has come to represent the entire DoD protocol suite based upon these protocols, including Telnet, FTP, UDP, and RDP. Technically, this is incorrect usage, because other protocol stacks can be layered on top of TCP and IP that provide similar services, but are not part of the DoD standard protocols (e.g., TP-4/IP, FTAM on TCP, etc.). Ideally, one should only use TCP/IP to mean the TCP protocol layered on top of the IP protocol. 


A commercial wide-area packet switching X.25 network. 


Telnet is a program that allows a computer user at one site to work on a computer at another site. It is the Internet standard protocol for remote terminal connection service. 

Telnet requires Internet access, that is, you must be on a TCP/IP network that gateways to the Internet. Unlike FTP and electronic mail, Telnet actually exposes you to the commands and programs of the remote host. 

For example, you can use the telnet command to run a program in your directory on a supercomputer hundreds of miles away. 


Technical/Office Protocol. A protocol stack for office automation developed by Boeing following the OSI model. This protocol is very similar to MAP except at the lowest levels, where it uses Ethernet (IEEE 802.3) rather than Token Bus (IEEE 802.4). 


The ISO protocol suite that performs the same functions as TCP/IP. TP-4 provides reliable, connection-oriented data streams using datagrams. This protocol also handles error detection, synchronization, and retransmission, just as TCP does. 


Time To Live. A field in a datagram designed to prevent packets from looping indefinitely in the Internet. Because routing information changes dynamically, two or more gateways may occasionally forward packets to each other in a loop, since each believes the other is the best route to the destination. A packet is initially sent with a nonzero TTL field, and each gateway that forwards that packet decrements the value in that field. Once the value reaches zero, a loop is assumed and the packet is discarded. 


User Datagram Protocol. The Internet standard protocol for sending datagrams between user programs. This protocol neither guarantees delivery nor does it require a connection. As a result it is lightweight and efficient, but requires the application to do all error processing and retransmissions. This protocol is built on top of IP and uses IP for datagram delivery (see TCP/IP). 


UNIX-to-UNIX-CoPy. This was initially a program run under the UNIX operating system (see BSD) that permitted one UNIX system to send files to another UNIX system via dial-up phone lines. Today, the term is more commonly used to describe the large international network made up of these machines using the UUCP protocol to pass netnews and electronic mail. 


A standard networking protocol suite approved by the CCITT and ISO. This protocol suite defines standard physical, link, and networking layers only (layers 1 through 3). X.25 networks are in use throughout the world. 


The CCITT standard for electronic mail. X.400 systems are in use in Europe, Canada, and several U.S. commercial installations. 


The CCITT standard for electronic mail directory services. 


Xerox Network Services. A proprietary networking architecture developed by Xerox.