SEECS Colloquium

Layered queueing is a kind of canonical form for extended queueing models, covering most forms of extended queueing models in an elegant way. The basic model is solved approximately in a conventional way. Various extensions have been found useful in modeling real systems, and require solution efforts that are specific to them. The layered model has also been extended or relaxed until it includes many forms of interaction that are not obviously layered. The interpretation of results from layered models, particularly bottleneck results, is special to this class of models. The seminar will describe the model, the extensions, and the interpretation, and say something about where layered modeling should be applied.

Murray Woodside holds the OCRI/NSERC Industrial Research Chair in Performance Engineering of Real-Time Software, and is Director of the Real-Time and Distributed Systems Lab (RADS Lab) in the Department. He is also Thrust Leader for the Software thrust of TRIO, the Telecommunications Research Institute of Ontario. He is presently Vice-Chairman of the Special Interest Group on Performance Measurement and Modelling (SIGMETRICS) of the ACM. Professor Woodside's recent research has concentrated on a special approach called "rendezvous nets" or "layered queueing", for modelling the performance limitations of distributed systems, that takes into account limitations due to the software design. This approach has been applied to client-server systems, distributed databases, software pipelines, transaction processing, communications switching software and (by others) object-oriented software in general, and to distributed systems based on DCE. Professor Woodside's research career began in modelling and control of stochastic industrial systems, and has included work on estimation and optimization in a broad range of systems. He also contributed to problems involving auto correlation in queueing systems, optimal allocation of work in queueing networks and decomposition methods for complex Petri Net models. Professor Woodside's consulting experience includes projects in building performance models, optimization, scheduling and simulation.