If we start a packet sniffer on a network segment with active RIPng router, most of the time we will only see the unsolicited Response messages similar to the one shown above. All routers send these updates with a period of 30 seconds, but independently of each other, i.e., with random phase shifts.

Note

Actually, for networks with many RIPng routers it is quite important to keep the routers unsynchronised, since otherwise a large number of Response messages sent within a short interval may adversely influence the network performance. Interestingly enough, weakly coupled periodic processes that are originally asynchronous often tend to synchronise. S. Floyd and V. Jacobson [FJ94] showed this can also happen to a set of routers exchanging periodic messages. To counter this phenomenon, the RIPng specifications requires the implementations either to derive the period of the messages from a precise clock that is not influenced by system load, or each time add a random offset to the nominal 30-second period so that the period is uniformly distributed between 15 and 45 seconds.

The essential part of the RIPng router operation involves the following two procedures:

1. Listen to the route advertisements sent by its neighbours and update its own routing table so that it contains the best routes (with shortest metrics) heard so far.

2. Every 30 seconds, send the current route table to all neighbours.