Self-Management: The Solution to Complexity or Just Another Problem?
IEEE Distributed Systems Online
Debugging debugged, a metaphysical manifesto of systems integration
ACM SIGSOFT Software Engineering Notes
SSS '08 Proceedings of the 10th International Symposium on Stabilization, Safety, and Security of Distributed Systems
SSS'06 Proceedings of the 8th international conference on Stabilization, safety, and security of distributed systems
A Lightweight Method for Automated Design of Convergence in Network Protocols
ACM Transactions on Autonomous and Adaptive Systems (TAAS) - Special Section: Extended Version of SASO 2011 Best Paper
A theory of fault recovery for component-based models
SSS'12 Proceedings of the 14th international conference on Stabilization, Safety, and Security of Distributed Systems
International Journal of Grid and Utility Computing
Self-stabilizing consensus average algorithm in distributed sensor networks
Transactions on Large-Scale Data- and Knowledge-centered systems IX
| Hi-index | 0.00 |
Objections to the practical use of stabilization have centered around problems of scale. Because of potential interferences between actions, global reasoning over the entire system is in general necessary. The complexity of this task increases dramatically as systems grow in size. Alternatives to dealing with this complexity focus on reset and composition. For reset, the problem is that any fault, no matter how minor, will cause a complete system reset with potentially significant lack of availability. For existing compositional alternatives, including compositional reset, severe restrictions on candidate systems are imposed.To address these issues, we give a framework for composition in which global reasoning and detailed system knowledge are not necessary, and which apply to a significantly wider range of systems than has hitherto been possible. In this framework, we explicitly identify for each component which other components it can corrupt. Additionally, the correction of one component often depends on the prior correction of one or more other components, constraining the order in which correction can take place. Given appropriate component stabilizers such as detectors and correctors, we offer several ways to coordinate system correction, depending on what is actually known about the corruption and correction relations.By reducing the design of and reasoning about stabilization to local activities involving each component and the neighbors with which it interacts, the framework is scalable. Reset isgenerally avoided by using the correction relation to check and correct only where necessary. By including both correction and corruption relations, the framework subsumes and extends other compositional approaches.Though not directly a part of this work, we mention tools and techniques that can be used to help calculate the dependency and corruption relations and to help create create the necessary stabilizers.To illustrate the theory, we show how this framework has been applied in our work in sensor networks.