The case for connecting a building
Wide area networks linked cities and countries, but most computer users in the 1970s did their work in one building or on one campus. A research group might have several minicomputers, a few terminals, a high-quality printer and a shared disk, all within a hundred yards of one another. Connecting them over telephone-style lines was overkill. What they needed was a way to join nearby machines cheaply and quickly. That need produced the local area network, usually shortened to LAN.
Before LANs, sharing resources in an office meant using a central computer with attached terminals or carrying data on disks and tape from machine to machine. Once personal computers and workstations began to multiply, buying a printer, a storage disk and a software license for every desk became expensive. Linking the machines promised to let people share the costly items.
Ethernet and other designs
Several local network designs appeared in the 1970s. The best known is Ethernet, developed at Xerox PARC in the early 1970s by a team that included Robert Metcalfe and David Boggs. In the original idea, all computers shared a single cable. Each computer listened before sending, waited if the cable was busy, and tried again after a random delay if two stations sent simultaneously and their messages collided. Metcalfe described the idea in a 1973 memo, and experimental versions ran at PARC, where the Alto computers and laser printers were being tried out. The same laboratory is discussed in the story of Xerox PARC.
Ethernet was not the only approach. Other designs included:
- Token Ring, which passed a special message called a token around a loop of stations so that only the holder could transmit. IBM promoted it in the 1980s.
- ARCnet, a token-based design from the late 1970s that was popular in small offices because it was inexpensive.
- Cambridge Ring and other university experiments in the United Kingdom.
- Proprietary systems from computer makers, such as networks that connected a company’s own machines and not much else.
A formal standards process under the Institute of Electrical and Electronics Engineers (IEEE) produced a family of local network standards in the early 1980s, and an Ethernet standard was published in 1983. Competing proposals were part of the debate. Over the following years, Ethernet’s simplicity, falling hardware costs and support from many manufacturers helped it become the usual choice, though it is hard to say with certainty why one technology wins over another.
Cables and hardware
The first Ethernet networks used thick coaxial cable, and later versions used thinner coaxial cable that was easier to install. A major change came in 1990 with a standard for running Ethernet over twisted-pair wire, similar to telephone wiring, connected through central devices called hubs. This allowed offices to reuse existing cable runs and made it easier to add or move a machine. Later devices called switches improved how traffic was handled by sending packets only to the port that needed them, which reduced collisions.
Each computer needed a network interface card, an expansion board that connected it to the cable. As prices fell in the late 1980s, adding a card became a routine office expense.
Software: servers, files and printers
Hardware alone was not much use without software. The most common early uses were simple and practical.
- File sharing. A file server, a computer with large disks, held documents and programs that many users could open from their own desks.
- Print sharing. One good printer could serve a whole department.
- Electronic mail inside the organization, covered more widely in the history of electronic mail.
- Shared databases, such as inventory or customer records, which became more common as business computing spread.
Companies such as Novell with its NetWare product in the 1980s, and Apple with AppleTalk, which was built into Macintosh computers from the mid-1980s, offered network software for personal computers. Microsoft added networking features to its operating systems over the years. The concept of clients, ordinary desktop machines, requesting services from servers became the dominant model for office computing.
Campuses and linking LANs together
As more buildings gained LANs, the next step was joining them. University campuses installed backbones, high-capacity cables running between buildings, and used bridges and routers to connect individual LANs. In many places, campus networks were then attached to wider academic networks, which is how a workstation in a dormitory could reach distant computers. The decision of how to link independent networks together using common rules is the subject of network architecture and internet protocols.
Wireless and what came later
Wireless local networks developed more slowly. A standard for them, IEEE 802.11, was first published in 1997, and the Wi-Fi name came into use in the following years. Over the 2000s, wireless networking in homes, cafes and offices became ordinary, supported by faster speeds and cheap hardware.
Effects on work
Local networks changed how organizations used computers. They made collaboration on shared documents possible, supported the growth of shared company databases, and encouraged the model of many small computers over one large central machine. This shift contributed to the long decline of the mainframe computers that had dominated earlier decades, though mainframes did not disappear and continue to handle certain tasks. It also introduced new tasks such as network administration, backups and the need to manage who could reach which files, bringing computer security into everyday office life.