THE STOCHASTIC RENDEZVOUS NETWORK MODEL FOR PERFORMANCE OF SYNCHRONOUS CLIENT-SERVER-LIKE DISTRIBUTED SOFTWARE

Distributed or parallel software with synchronous communication via re ndezvous is found in client-server systems and in proposed Open Distri buted Systems, in implementation environments such as Ada, V, Remote P rocedure Call systems, in Transputer systems, and in specification tec hniques such as CSP, CCS and LOTOS. The delays induced by rendezvous c an cause serious performance problems, which are not easy to estimate using conventional models which focus on hardware contention, or on a restricted view of the parallelism which ignores implementation constr aints. Stochastic Rendezvous Networks are queueing networks of a new t ype which have been proposed as a modelling framework for these system s. They incorporate the two key phenomena of included service and a se cond phase of service. This paper extends the model to also incorporat e different services or entries associated with each task. Approximati ons to arrival-instant probabilities are employed with a Mean-Value An alysis framework, to give approximate performance estimates. The metho d has been applied to moderately large industrial software systems.