Layered interactive convergence for distributed clock synchronization

A clock synchronization service ensures that spatially dispersed and heterogeneous processors in a distributed system share a common notion of time. In order to behave as a single, unified computing resource, distributed systems have need for a fault-tolerant, clock synchronization service. One approach employs the interactive convergence (ICV) method, which is both fully distributed and inherently fault-tolerant. However, this approach suffers from limits in terms of system scalability due to resource overhead. In this paper, a new layered form of ICV is introduced, compared with basic ICV and other simpler alternatives via experimental analysis on a distributed system testbed, and shown to improve synchronization tightness, resource utilization, and scalability. Index Terms – Cluster computing, interactive convergence, clock synchronization, convergence, fault tolerance.