Skip to Main Content
Distributed application management system is important for managing applications on distributed computing platforms. One of the main caveat of using a distributed management system is that the management system itself, as a distributed application, need to be deployed and maintained continually. In this paper, we propose Self-Managed Overlay Network (SMOM) and explore the challenges associated with designing a management system with self-management capability. SMON manages itself using epidemic approach at runtime. SMON can automatically deploys itself to a set of machines and recovers failed peers securely. It can also upgrade itself to new versions online. Through mathematical analysis and evaluation on Planet-Lab platform, we show that SMON achieves good performance and scalability.