NetQoPE: Middleware-based Network QoS Provisioning Engine for Distributed Real-time and Embedded Systems

Standards-based quality of service (QoS)-enabled component middleware is increasingly used as a platform for networkcentric, distributed, real-time and embedded (DRE) systems where applications require ensured network QoS even when network resources are scarce or congestion is present. Traditional approaches to building network-based, QoS-sensitive applications, such as Internet telephony and streaming multimedia applications, have relied on leveraging low-level proprietary command line interfaces such as CISCO IOS and/or low-level management interfaces such as SNMP. Although QoS-enabled component middleware offers many desirable features, until recently it has lacked the ability to simplify and coordinate application-level services to leverage advances in end-to-end network QoS management. Our work models and provisions network QoS in QoS-enabled component middleware for DRE systems, extending our previous work that predominantly used host resources to provision end-system QoS. By integrating a Bandwidth Broker (BB) that provides network QoS into our DRE middleware, we enable end-to-end QoS assurance. We describe a declarative QoS provisioning solution called NetQoPE. Its modeling capabilities allow a user to specify application QoS requirements in terms of network and computational needs. The model, in turn, leads to deployment decisions that take into account network resources (in addition to host resources). Finally, our DRE middleware provisions network and middleware elements to enable and enforce QoS decisions. We demonstrate and evaluate the effectiveness of NetQoPE in the context of a representative DRE system – a modern enterprise environment with a single Layer-3/Layer-2 network that supports different types of traffic. Our empirical results show that the capabilities provided by NetQoPE yield a predictable and efficient system for QoS sensitive applications even in the face of changing workloads and resource availability in the underlying network subsystem.