Replicated Data Management for Mobile Computing (Synthesis Lectures on Mobile & Pervasive Computing)

Managing data in a mobile computing environment invariably involves caching or replication. In many cases, a mobile device has access only to data that is stored locally, and much of that data arrives via replication from other devices, PCs, and services. Given portable devices with limited resources, weak or intermittent connectivity, and security vulnerabilities, data replication serves to increase availability, reduce communication costs, foster sharing, and enhance survivability of critical information. Mobile systems have employed a variety of distributed architectures from client–server caching to peer-to-peer replication. Such systems generally provide weak consistency models in which read and update operations can be performed at any replica without coordination with other devices. The design of a replication protocol then centers on issues of how to record, propagate, order, and filter updates. Some protocols utilize operation logs, whereas others replicate state. Systems might provide best-effort delivery, using gossip protocols or multicast, or guarantee eventual consistency for arbitrary communication patterns, using recently developed pairwise, knowledge-driven protocols. Additionally, systems must detect and resolve the conflicts that arise from concurrent updates using techniques ranging from version vectors to read–write dependency checks. This lecture explores the choices faced in designing a replication protocol, with particular emphasis on meeting the needs of mobile applications. It presents the inherent trade-offs and implicit assumptions in alternative designs. The discussion is grounded by including case studies of research and commercial systems including Coda, Ficus, Bayou, Sybase’s iAnywhere, and Microsoft’s Sync Framework. Table of Contents: Introduction / System Models / Data Consistency / Replicated Data Protocols / Partial Replication / Conflict Management / Case Studies / Conclusions / Bibliography

Douglas B. Terry is a principal researcher at the Microsoft Research Silicon Valley laboratory. His research focuses on the design and implementation of novel distributed systems and addresses issues such as information management, fault-tolerance, and mobility. He currently is serving as chair of ACM's Special Interest Group on Operating Systems (SIGOPS). Before joining Microsoft, Doug was the cofounder and CTO of Cogenia, chief scientist of the Computer Science Laboratory at Xerox PARC, and an adjunct professor in the Computer Science Division at UC Berkeley, where he regularly teaches a graduate course on distributed systems. Doug has a PhD in computer science from UC Berkeley.