A Technical History of National Physical Laboratories
Architecture - A Technical Tour
Conventional, point-to-point communications networks like
telephone networks had a reliability problem since they would lose the
connection if any one link failed. And because there were necessarily
more links in longer distance communications, long distance communications
were proportionately less reliable and more expensive.
Donald Davies' proposed network architecture solves this
problem by dividing each communication into individual, digital packets,
and then using routing protocols to get the packets from one place to
another. Each message is divided into many individual packets of information,
and then each packet is sent to its destination independently. The network
consists of nodes (similar to routers today) connected by digital links.
The network carries short messages which are stored at each node through
which they pass (called store-and-forward or message-switching systems).
Each message enters the network in a well-defined format and includes
some routing data. The responsibility for putting messages into this format
belongs to the network, not the user. Between the network and the users
are interface computers (similar to hosts or end systems today) each handling
a mixed collection of subscribers in a geographical area. In Davies' architecture,
time-division multiplexing would allow multiple users to take turns transmitting
portions of their messages. If a user had a short message, such as a single
command, the whole message could be sent in the first packet, while longer
messages would take several time slots to transmit.
Packet switching, in Davies' view, was the communications
equivalent of time-sharing: it would maximize access to a scarce resource
in order to provide interactive computing. Davies thought the pressing
need was for a network that could serve the users of commercial time-sharing
services, and also remain simple and easy to use. His network architecture
served the aim of building a commercial system mainly by bringing down
the cost of data communications. Ultimately, Davies thought that his architecture
using packet switching technology could be a commercial product that would
contribute directly to the Ministry of Technology's plan to revitalize
the British economy.
Figure 1 (Davies)
The network is broken into two sections: a high-level network called
the subnet and a local network built around the interface computer.(Campbell