AIX > Administrator > Networks

Communications Protocols 101

This is the first installment in a series of articles on TCP/IP communications. Read part two and part three.

Connecting two or more computers to form a network is a common practice today. The original computer network, ARPANET, was developed in the 1960s under the sponsorship of the United States Department of Defense. ARPANET was a packet-switched network, which meant that messages transmitted between computers were broken into packets. These packets had a maximum size and were encapsulated with header information that provided addressing for routing and reassembling at the destination. The term Internet was coined in 1983 when ARPANET was split into two pieces--MILNET (for the military) and a reduced version of ARPANET. The building blocks of the Internet are its (networking) IP and its process-to-process TCP used through physical entities of WANs and LANs. The distinction between a WAN and a LAN is typically the size of the geographical area. WANs can span cities, countries or even continents. LANs are typically limited to a building or a business/university campus.

I'll start with a brief explanation of a network, including Ethernet and Token Ring basics, followed by concepts of Open Systems Interconnection (OSI) and Internet models. Next, I'll introduce TCP, User Datagram Protocol (UDP) and IP protocols (a set of conventions governing data treatment and formatting in an electronic communications system), including communications ports and sockets.

Network Basics

Let's start with a view of a simple network. Figure 1 shows two computers communicating over a network. The end system (e.g., a customer) is considered the ultimate producer or consumer of network information. An end-to-end protocol (e.g., TCP) operates between two end systems.

A hop-by-hop protocol (e.g., IP) operates between an end system and one or more intermediate systems (e.g., router) to reach a destination end system. A point-to-point protocol operates on a shared physical media (e.g., Ethernet) connecting two systems. A router is considered as an intermediate system that's multi-homed--a system with attachments to two or more physical networks. A router's job is to receive data on an interface destined for a different network to which it's attached, and then forward the data to that network. This action is typically called store and forward.

A bus topology is a physical network architecture in which all hosts contend (potentially simultaneously) for access to the same media (wire or cable, sometimes called a cable plant). A ring topology (e.g., IBM's Token Ring) is a physical network architecture in which hosts must receive permission to use a shared media by receiving a token, a special frame granting permission to transmit.

Ethernet is a multi-access, packet-switched communications system for carrying digital data among locally distributed computing systems. In a packet-switched network, data flows through individual units or blocks of information. Packets contain both headers (control and routing information) and data (end-user data or control data). In an Ethernet LAN system, all computers connect to a shared coaxial cable through a network interface card (NIC). Also in an Ethernet LAN system, the control mechanism is an access method known as Carrier Sense Multiple Access/Collision Detect (CSMA/CD). To contend for the use of the cable plant, any NIC wishing to transmit checks for a busy signal (using its carrier sense capability). If the cable plant hasn't been busy for a specified amount of time, the station begins to immediately transmit its data.

Tom Farwell is a technical editor for IBM Systems Magazine, Open Systems edition. He can be reached through

comments powered by Disqus



2019 Solutions Edition

A Comprehensive Online Buyer's Guide to Solutions, Services and Education.


A Primer on Power Systems 10 Gb Ethernet

IBM Systems Magazine Subscribe Box Read Now Link Subscribe Now Link iPad App Google Play Store
IBMi News Sign Up Today! Past News Letters