Disruption-aware service composition and recovery in dynamic networking environments

The dynamic and heterogeneous natures of large-scale systems pose fundamental challenges to the design of service composition methods with minimum service disruptions. Improving reliability has long been a topic of extensive research in large-scale systems. Little existing work, however, has considered service deliveries spanning multiple components and taken both failure duration and frequency into account. This paper proposes a new service composition and recovery framework designed to achieve minimum service disruptions. The framework consists of two-tiers: service routing, which selects the service components, and network routing, which finds the network path that connects these service components. Our framework is based on a novel concept: disruption index, which characterizes different aspects of service disruptions, including frequency and duration. We formulate the problem of minimum-disruption service composition and recovery (MDSCR) as a dynamic programming problem and give its optimal solution under the assumption of complete knowledge of future failure distribution. We then present our MDSCR heuristic, which approximates the optimal solution with one-step lookahead prediction, where service link lifetime is predicted through statistical regression. We present the preliminary performance results of our algorithm via simulation study.