An Overview of PANACEA, a Software Package for Analyzing Markovian Queueing Networks

Abstract

PANACEA is a software package that significantly extends the range of Markovian queueing networks that are computationally tractable. It solves multi-class closed, open, and mixed queueing networks. Based on an underlying theory of integral representations and asymptotic expansions, PANACEA solves queueing networks that are orders of magnitude larger than can be solved by other established algorithms. The package is finding widespread use in Bell Laboratories. It also has important software innovations. A flexible programming-language-like interface facilitates compact representation of large queueing networks. An out-of-core implemental strategy enables PANACEA to be ported to processors with modest memory. The modular structure of this software package, along with the automatic machine-generated parser, makes it easily extendable. This paper provides an overview of two basic versions of PANACEA, versions 1.0 and 1.1, which solve “closed” networks only. A description of its model language is given from the point of view of its capability to describe queueing networks in a compact, natural manner. The paper discusses the algorithms, together with their time and storage requirements, that are used in the implementation. Several numerical examples are given.