A hierarchical asynchronous replication protocol for large scale systems

Abstract

The authors present a new asynchronous replication protocol that is especially suitable for wide area and mobile systems, and allows reads and writes to occur at any replica. Updates reach other replicas using a propagation scheme based on nodes organized into a logical hierarchy. The hierarchical structure enables the scheme to scale well for thousands of replicas, while ensuring reliable delivery. A new service interface is proposed that provides different levels of asynchrony, allowing strong consistency and weak consistency to be integrated into the same framework. Further, the scheme provides the ability to offer different levels of staleness, depending upon the needs of various applications, by querying from different levels of the hierarchy. Also, it allows a selection from a number of reconciliation techniques based on delivery order mechanisms. Restructuring operations are provided to build and reconfigure the hierarchy dynamically without disturbing normal operation. The scheme tolerates transmission failures, node failures and network partitions.