Point-to-Point Protocol over Ethernet (PPPoE) allows PPP to run over Ethernet. Typically, PPP runs over serial interfaces and in most cases runs over phone lines connected to a server.
With DSL and cable modems, where a personal computer is connected to the Ethernet interface of the modem, ISPs cannot run PPP because PPP cannot run directly over Ethernet. ISPs often prefer to use PPP to provide features such as user authentication and bandwidth monitoring.
Typically, PPPoE is set up in two different configurations: PPPoE for the single user (Figure 15) or PPPoE on a local network. For locations with single computers, the PPPoE client is typically loaded on the computer and it reaches the PPPoE server through the Ethernet connection via the DSL modem. The DSL modem then forwards the packets to the WAN interface without interpreting the PPPoE packets. The PPPoE packets reach their final destination (PPPoE server) for further handling. This implementation is in compliance with RFC 2516.
Figure 15 PPPoE for single user
The second configuration is usually seen in multi-computer locations, small offices, or branch offices where the entire LAN is connected to the Internet via DSL or cable modem. In this case, either the modem or the gateway acts as a PPPoE client. Figure 16 shows how a Contivity 1100 gateway connected to the DSL modem acts as the PPPoE client. In this configuration, the PPPoE client encapsulates the LAN traffic in the PPPoE header and forwards it to the PPPoE server.
Figure 16 PPPoE on a local network
PPPoE has the following usage restrictions: