Distributed Arrays: A P2P Data Structure for Efficient Logical Arrays

Distributed hash tables (DHT) are used for data management in P2P environments. However, since most hash functions ignore relations between items, DHTs are not efficient for operations on related items. In this paper, we modify a DHT into a distributed array (DA) that enables efficient operations on logical arrays. The array elements of a DA are placed in a P2P overlay network according to a simple rule such that the load is balanced and the number of messages required to access elements sequentially is reduced. The number of messages required for array operations is much smaller than that for operations on DHTs. We demonstrate this theoretically and experimentally.