A Randomized ID Selection Algorithm for Peer-to-Peer Networks

We present a low-cost, decentralized algorithm for the problem of selecting host IDs in peer-to-peer systems. The problem arises in the context of distributed hash tables (DHTs). Our algorithm is the first one with the following properties: (a) both arrivals and departures of hosts are handled, (b) partition balance ratio is guaranteed to be at most 4, (c) at most one existing ID is re-assigned in response to departures, and (d) the expected cost is Θ(R + logn) messages, where n denotes the current number of participants, and R denotes the cost of routing one message in the DHT. Further, our algorithm is independent of the overlay routing topology, making it applicable to all DHTs. Our algorithm improves upon existing algorithms which have one or more of the following drawbacks: they require Θ(R logn) messages, do not handle host departures, or re-assign the IDs of O(log n) existing hosts in response to a departure. Our ID selection algorithm enables emulation of a variety of deterministic and randomized families of interconnection topologies, in a straightforward fashion. Finally, an extension to our algorithm allows improvement of the partition balance ratio to (1 + ), where > 0, albeit at the cost of re-assigning the IDs of O( ) hosts upon arrival/departure. We believe that ours is the first algorithm that allows such fine-tuning.