Abstract: In an ATM network operating in accordance with PNNI standards, the operation of a switch is divided into two phases, a learning mode phase and an active mode phase. During learning mode phase, each time a reachability is received from a sending switch, the reachability is recorded in the database of the receiving switch and the received reachability is sent to other switches in the same peer group. At the end of the learning mode phase, the database of each of the switches contains the reachabilities advertised by all the switches of the peer group during the learning mode group phase. On a termination of the learning mode, the switches enter the active mode phase. During the active mode, each time a reachability is received from a sending switch, the receiving switch checks to see if the reachability has been advertised in the past by that sending switch. If it has, the reachability is considered to be valid and the receiving switch waits for the next reachability.

Abstract: A leaky bucket checker which combines synchronous updates with event driven asynchronous updates triggered by packet arrivals. A synchronous update is performed exactly as an event-driven update by assuming that a packet of length zero has arrived at the time of the update. These updates are performed in a round robin fashion on each connection. Therefore, assuming that one such update can be performed in each clock tick, the maximum bit representation of the last update time state variable can be limited to N bits for 2.sup.N total connections. Given the reduced processing and storage costs, a great number of network connections and leaky bucket checkers is possible.