Special issue on tools for computer performance modeling and reliability analysis

Giuliano Casale, Richard R. Muntz, Giuseppe Serazzi
2009 Performance Evaluation Review  
This special issue offers an up-to-date overview on software tools for computer performance modelling and reliability analysis that are currently developed by research groups in academia. Because of the ever increasing complexity of computer systems, researchers have developed throughout the years many analytical and simulative approaches to evaluate performance and reliability of realworld systems. However, the mathematical skills required to understand some of these results are often a
more » ... for their widespread adoption as a part of practitioners tend to perceive research results on modeling as too complex to be useful in their problems. Software tools represent a notable exception to this barrier, as they tend to hide most of the mathematical details of the modeling or analytical technique. Further, the learning curve of users can be made faster by distributing examples or graphical interfaces with context-sensitive hints. Although software tools represent an effective way to convey modeling research result into practical applications, we feel that more should be done in advertising to the general public what tools are available and how they differ in terms of functionalities and modeling approaches used. For example, we are not aware of any web page listing the available academic performance and reliability tools. Therefore, the chances of a tool of becoming popular seems to be left either to the individual action of the authors or to the popularity of the web platform on which the tool is distributed (e.g., Sourceforge). The eleven invited papers in this special issue try to help in this direction by reviewing the general characteristics of some popular tools for performance and reliability analysis. Each paper presents a general description of tool features together with a discussion on upcoming extensions. . The tool is specifically tailored to the solution of Generalized Stochastic Petri Nets (GSPNs) and Stochastic Well-Formed Nets (SWNs) and can perform a variety of evaluations such as structural analyses, statespace analysis, simulation, Markov chain generation. Java Modelling Tools (JMT) is a suite of applications for queueing network analysis and workload characterization. The tool is presented in the paper "JMT Performance Engineering Tools for System Modeling" authored by Marco Bertoli, Giuliano Casale, and Giuseppe Serazzi. JMT supports product-form and non-productform networks, what-if analysis, wizard-based user interaction, and a module for workload characterization from measured traces. , is a multi-paradigm environment that allows the specification of composite performance and reliability models based on different formalisms such as stochastic activity networks (SANs), fault trees, and the PEPA stochastic process algebra. Möbius simulation traces can be exported in XML to the Traviando visualization tool for model validation and statistical analysis. OPEDo is a software tool for the numerical optimization of performance or dependability metrics in discrete event systems. The application is described in the paper "OPEDo: A Tool for the Optimization of Performance and Dependability Models", by Markus Arns, Peter Buchholz, and Dennis Müller. OPEDo implements local, global, and hybrid search methods which iteratively evaluate the response surface seeking for the optimum. OPEDo considers the underlying model as black-box, thus it supports the interfacing to external applications for model specification and evaluation. PEPA Eclipse Plug-in, presented in the paper "The PEPA Eclipse Plugin" by Mirco Tribastone, Adam Duguid, and Stephen Gilmore, supports the definition of stochastic process algebra models based on the Performance Evaluation Process Algebra (PEPA) language.
doi:10.1145/1530873.1530875 fatcat:ou4f6oof4zc25ieie5kavcykmu