UNIX DHCP Historical background

DHCP Historical background

At first, most TCP/IP networks were relatively small and static. Manual IP address management techniques were sufficient for them. Each station kept its own IP address somewhere in its secondary storage. Once the address had to be changed, it required manual administrator action, usually at the machine console, and in most cases involved a reboot.

Soon afterwards, as more complex networks were established, as more and more underlying network hardware was used for TCP/IP communication networks and as cheap client workstations without secondary storage came in use, a need for central administration of the hardware to IP addresses bindings became understandable. A special protocol (RARP) for such bindings was designed. It allowed a machine on a network segment to learn its own IP address and then to begin normal TCP/IP operation.

Another protocol, BOOTP, was also developed to allow diskless stations retrieve all the TCP/IP configuration parameters and other operating system data, needed to start functioning normally after a startup.

The next extension to BOOTP provided the Dynamic Host Configuration Protocol, DHCP. There are two primary differences between DHCP and BOOTP. First, DHCP defines mechanisms through which clients can be assigned a network address for a finite lease, allowing for serial reassignment of network addresses to different clients. Second, DHCP provides the mechanism for a client to acquire all of the IP configuration parameters that it needs in order to operate. So DHCP is based on BOOTP, adding the capability of automatic allocation of reusable network addresses and additional configuration options. DHCP captures the behavior of BOOTP relay agents, and DHCP participants can interoperate with BOOTP participants.