Large-Scale Weakly Consistent Replication using Multicast

In today's Internet, there exist several repositories of resource allocation information. Specifically, these registries contain information about IP address space delegations, name space allocations and inter-ISP routing policies. Such registries are useful for coordinating allocation of Internet names and addresses, and for debugging network routing. For performance and availability reasons, there is an increasing need to replicate these registries Internet-wide. This paper describes the design of an weakly consistent replication scheme that uses IP multicast. The IP multicast service is unreliable; we use a stochastic wait with suppression technique to scalably recover lost updates. This technique reduces duplicate retransmissions of lost updates. Detailed simulations demonstrate that, even in large sparsely-distributed registries, less than 10% of the losses result in duplicate retransmissions. For the expected update traffic patterns, our approach outperforms other multicast loss r..