Subrange Caching: Handling Popular Range Queries in DHTs

Range query, which is defined as retrieving all the keys within a certain range over the underlying DHT, has attracted a lot of attention recently. However, little has been done to speed up range queries through caching, one of the least studied P2P problems. In this paper, we present a caching protocol that (1) has excellent parallelizability, (2) can achieve O(1) complexity for moderate range queries, and (3) balances the access load among all the DHT peers. Our experimental results shows significant gains in both range query performance and load balancing, with minimal overhead.