Self-Adaptation of DHT Routing Table Size with Attractor Selection

摘要

In this paper we discuss the self-adaptation of the routing table size of a distributed hash table (DHT) by applying the biologically inspired attractor selection method to adapt to variations in the traffic caused by uncontrollable and unpredictable fluctuations in the underlay network. Since the common DHT mechanisms provide only static settings of the routing table, our goal is to perform an adaptive control, as unsuitable parameter settings would cause too much overhead traffic and deteriorate the overall network performance. We adopt attractor selection as adaptation scheme, since it provides better self-adaptablity and robustness features than other methods and we demonstrate its applicability to control the DHT node state for the case of the routing table size of Pastry, a well-known DHT algorithm. Our goal is to minimize the overhead traffic when the the conditions in the underlay network change.