ARP/RARP PCI
3 - [Reverse] Address Resolution Protocol
Move the mouse pointer to any protocol field and press the left mouse button.
Both ARP and RARP are address resolution protols (RARP is a variant of ARP). The are needed because IP uses logical host addresses (IP addresses) while media access control (Ethernet, Token-Ring, FDDI etc.) needs MAC addresses.
"The world is a jungle in general, and the networking game contributes many animals." [RFC 826]
IP addresses are assigned by network management (network administrator) to IP hosts. This is usually done by configuration file options and driver software (e.g. ifconfig command on UNIX machines).
Since LAN topologies (Ethernet, Token-Ring, FDDI etc.) can't use IP addresses, the IP addresses must be mapped to the corresponding MAC addresses.
The first implementation of TCP/IP protocols on a LAN basis was done on UNIX machines at the University of California, Berkeley, USA, which used a mapping table called /etc/ethers ; this was an statically assignment. But as soon as the TCP/IP networks grew up it became impossible to keep /etc/ethers files up to date. A dynamic assignment procedure for mapping had to be implemented: ARP / RARP.
1. ARP
Searched: Another host's MAC address:
If an IP hosts want to call another IP host knowing only the IP address, an ARP request via LAN broadcast is sent to all local IP hosts. The called station (if active) will answer with an ARP reply telling its MAC address in return.
2. RARP
Searched: The local machine's IP address:
Diskless workstations cannot read their own IP address from configuration files (because the haven't any). They send an RARP request (or BOOTP request) to a RARP server (or BOOTP server); the RARP server will find the corresponding IP address in its configuration files (by means of the requesting station's MAC address), sending this IP address back with a RARP reply packet.
Literature/Sources: