Skip to Main Content
This paper proposes a feedback-loop infrastructure attached to our legacy multimedia streaming system to continually monitor its activities and performance and rapidly recover the system from failures or error states. Our infrastructure consists of four components: monitors with sensors to measure and record the activities and state of among the components of the running system, analyzer to interpret and abstract the logged data, planner to decide what action to take for reported situations and executants to adjust and reconfigure the system via effectors. An event-based schema is proposed to efficiently deliver recorded system information among our infrastructure components. We summarize our successful examples using this approach in dynamically self-healing for our streaming application.