Approximate MVA from Markov model of software client/server systems

Stochastic rendezvous networks (SRVN) are performance models for multitasking parallel software with intertask communication via rendezvous (RV). The paper describes an approximation for the arrival instant probabilities for a class of simple client/server SRVN, based on a rigorous analysis of the Markov chain model describing the interference of different client tasks contending for a single server task with FIFO queueing discipline. The algorithm for a simple client/server SRVN is integrated into an iterative decomposition algorithm for complex SRVN models with any number of servers and multi-layered service. Finally, the SRVN model is used to study an example of software bottleneck.<
>