Generic Self-Healing via Rejuvenation: Challenges, Status Quo, and Solutions

摘要

Software rejuvenation - in its simplest form a restart of a component or a program - is an efficient and universal approach for ad hoc healing of certain complex systems such as SOA components, telecommunication systems, and servers in data centers. Despite of its advantages this technique has not been widely deployed in other scenarios. The reasons are several shortcomings including loss of application availability and loss of working data due to a restart, and a lack of standardized support in operating systems, middleware, and component frameworks. In this position paper we argue that even partial remedies to these problems can turn rejuvenation into a powerful self-healing tool applicable to a larger variety of scenarios. We discuss rejuvenation-related problems, overview existing solutions, and propose a set of efficient architectural approaches which can pave the way to a universal adoption of this technique.