Representing the explicit state space of performance models has inherent difficulties. Just as the state-space explosion effects functional correctness evaluation, so it can also be easily a problem in performance models. In particular, classical Markov chain analysis of any variety requires exploration of the global state space and, even for a simple system, this quickly becomes computationally infeasible. Fluid and mean-field analysis techniques attempt to side-step the state-space explosion and provide a computationally cheap way of analysing certain features of Markov chains.
Invited talk for QAPL 2012
pubs.doc.ic.ac.uk: built & maintained by Ashok Argent-Katwala.