Quality of Service Management of Distributed Realtime Embedded Systems

With increasing advantages of cost, performance, and scale over single computers, the proliferation of distributed systems in general and distributed event-based systems in particular have increased dramatically in recent years. To support the requirements of a broad spectrum of application domains, publish/subscribe middleware for event-based distributed systems typically provides many policies that affect end-to-end system quality of service (QoS) properties. While tunable policies provide fine-grained control of system QoS, configuring and managing QoS has become more complex as QoS support for pub/sub middleware platforms has increased. This work focuses on the following four synergistic areas: (1) a model-driven technique for developing QoS configurations and applicable implementation artifacts; (2) composite QoS metrics to quantitatively evaluate multiple QoS concerns, (3) a technique for evaluating QoS mechanisms in specific operating environments and providing guidance for QoS configurations; and (4) a technique to support autonomic configuration and adaptation of QoS-enabled pub/sub middleware in flexible and dynamic environments.