Group Anti-Entropy - Achieving Eventual Consistency in Mobile Service Environments

Abstract

Data consistency protocols are vital ingredients of mobile data management systems. Notable research efforts have been spent to find adequate consistency models for allowing mobile and nomadic users to share mutable data. Recently, mobile Ambient Service infrastructures that pose somewhat different requirements have entered the focus of attention. Ambient services are not as loosely coupled as the afore-mentioned systems, but they still need flexible consistency protocols that may adapt to the current dynamics in the system. We propose an extension to the well-known anti-entropy protocol that makes use of the nature of Ambient Service environments to allow for a flexible consistency management among arbitrary groups of mobile service replicas. We will show that our protocol can exploit the concept of group updates to increase its efficiency in terms of bandwidth usage. Furthermore, we prove that it avoids costly state transfers by means of a simple rule that limits the divergence within the overall set of replicas. Finally, we introduce two parameters for adjusting the level of consistency in system, and we present experimental results that show the effectiveness and the efficiency of the protocol.