Self-stabilizing Middleware Services

Abstract

We present a set of middleware services with the self-stabilization fault-tolerance property, which automatically returns a system to its intended behavior when transient faults corrupt the system's state. We first propose a self-stabilizing virtual synchrony service realizing state machine replication in a setting with a fixed set of processors. We then develop a multipurpose reconfiguration service and suggest how the virtual synchrony service can be deployed in settings with more dynamic processor sets. This currently provides the theoretical basis to move into prototyping and evaluation, and can possibly lead to self-stabilizing versions of important services like Chubby, Zookeeper and Vsync.