A scalable fault tolerant approach to core election in an inter-domain multicast routing environment

We propose a consensus protocol to rearrange the multicast core based tree, after a core failure. However, the applications of this protocol are not limited to core failure. It can be used in any situation where a common value should be fixed by consensus in an inter-domain routing environment. This protocol is a hierarchical (two-level) extension of the original proposal by Chandra and Toueg (1996). However, in order to use the algorithm in an inter-domain environment such as the Internet, we introduce randomization in our consensus protocol. Indeed, we minimise by this way the total number of rounds. A comparison with other consensus protocols shows that our solution gives a better complexity of the algorithm. This solution also opens new perspectives in terms of core election in an inter-domain multicast routing environment, since the classical approaches do not scale, or are based on manual configurations