Skip to Main Content
A distributed hash table (DHT) with replicated objects enjoys improved performance and fault-tolerance but calls for effective replica management. This paper deals with proximity-aware distributed mutual exclusion (PADME) for P2P replica management on a DHT. Three main components are involved in PADME: (1) a few nodes designated as the sink candidates for collecting and consolidating replica updates, (2) a node selected from sink candidates to execute gathered replica updates, and (3) a proximity-sorted replica list to guide propagating the updated result effectively and reliably across all replica holders. Simulation results demonstrate that PADME exhibits at least two orders of magnitude less update message traffic than known leading distributed mutual exclusion-based algorithms for DHT replica management (namely, Sigma and E2E) under various cases examined. As a result, PADME outperforms Sigma (or E2E) by an order of magnitude (or up to 50%) in terms of the update throughput, while drastically lowering its update latency by up to 3 orders (or an order) of magnitude.