Runtime monitoring of distributed systems

Abstract

Distributed and component-based architectures are becoming more prevalent computer systems. The increased complexities introduced by the distribution hampers dependability, emphasising the need for verification techniques tailored for a distributed setting. Runtime verification has proven to be a viable approach for verifying correctness, by focussing on the adherence of the runtime-generated trace to the desired properties. We present a broad taxonomy of current techniques to distributed monitoring, culminating in the proposal of a novel migrating monitor approach. We argue for certain situations where this approach presents clear advantages over current techniques.