Adaptive Self-Configuration of EJB Server: A Performance Model Approach

Abstract

High dynamic computing environment makes quality of service (QoS) guarantee more important to distributed system. It should possess self-tuning capability reacting to external environment variation such as fluctuating workload. This paper proposed an adaptive self- configuration framework for EJB platform, which can automatically tune configuration parameters to preserve QoS when workload changed. The key point of this framework is a performance model based on layered queuing network, which guides the search for the best combination of configuration parameters to satisfy performance requirement of EJB application. We have prototyped this self-configuration framework on an EJB server named OnceAS, and the experiment results showed that through the framework's control, system performs well on QoS goal.