An Integrated Approach for Quality of Service in Cluster Networks

Multiprocessor interconnects have been designed primarily to provide high throughput and low latency for best-effort traffic. This is true for processing node architectures and communication software as well. However, with the increasing use of these systems (clusters in particular) for interactive applications such as multimedia, web servers and electronic commerce, predictable communication performance, called Quality of service (QoS), has become a major concern. Many of these applications require guaranteed communication delay and bandwidth, in addition to low latency and high throughput. The main focus of this research is to design an end-to-end communication infrastructure for a cluster system to support different applications with varying QoS requirements. The research activities include (1) router architecture design and evaluation to provide predictable and high performance; (2) network interface and software messaging layer support to inject/eject traffic into/from the network as per QoS requirements; and (3) CPU scheduling mechanisms that propagate these capabilities up to the application level. The research proposes to seamlessly integrate these three components to provide a high performance communication architecture with QoS guarantees.