Using genetic search for reverse engineering of parametric behavior models for performance prediction (bibtex)
by Krogmann, Klaus, Kuperberg, Michael and Reussner, Ralf
Abstract:
In component-based software engineering, existing components are often re-used in new applications. Correspondingly, the response time of an entire component-based application can be predicted from the execution durations of individual component services. These execution durations depend on the runtime behaviour of a component, which itself is influenced by three factors: the execution platform, the usage profile, and the component wiring. To cover all relevant combinations of these influencing factors, conventional prediction of response times requires repeated deployment and measurements of component services for all such combi- nations, incurring a substantial effort. This paper presents a novel comprehensive approach for reverse engineering and performance prediction of components. In it, genetic programming is utilised for reconstructing a behaviour model from monitoring data, runtime bytecode counts and static bytecode analysis. The resulting behaviour model is parametrised over all three performance-influencing factors, which are specified separately. This results in significantly fewer measurements: the behaviour model is reconstructed only once per component service, and one application-independent bytecode benchmark run is sufficient to characterise an execution platform. To predict the execution durations for a concrete platform, our approach combines the behaviour model with platform-specific benchmarking results.We validate our approach by predicting the performance of a file sharing application.
Reference:
Using genetic search for reverse engineering of parametric behavior models for performance prediction (Krogmann, Klaus, Kuperberg, Michael and Reussner, Ralf), In IEEE Transactions on Software Engineering (Harman, Mark, Mansouri, Afshin, eds.), IEEE, volume 36, 2010.
Bibtex Entry:
@article{krogmann2009c,
abstract = {In component-based software engineering, existing components are often re-used in new applications. Correspondingly, the response time of an entire component-based application can be predicted from the execution durations of individual component services. These execution durations depend on the runtime behaviour of a component, which itself is influenced by three factors: the execution platform, the usage profile, and the component wiring. To cover all relevant combinations of these influencing factors, conventional prediction of response times requires repeated deployment and measurements of component services for all such combi- nations, incurring a substantial effort. This paper presents a novel comprehensive approach for reverse engineering and performance prediction of components. In it, genetic programming is utilised for reconstructing a behaviour model from monitoring data, runtime bytecode counts and static bytecode analysis. The resulting behaviour model is parametrised over all three performance-influencing factors, which are specified separately. This results in significantly fewer measurements: the behaviour model is reconstructed only once per component service, and one application-independent bytecode benchmark run is sufficient to characterise an execution platform. To predict the execution durations for a concrete platform, our approach combines the behaviour model with platform-specific benchmarking results.We validate our approach by predicting the performance of a file sharing application.},
author = {Krogmann, Klaus and Kuperberg, Michael and Reussner, Ralf},
doi = {10.1109/TSE.2010.69},
editor = {Harman, Mark and Mansouri, Afshin},
isbn = {00985589 (ISSN)},
issn = {00985589},
journal = {IEEE Transactions on Software Engineering},
keywords = {Genetic search,advert_pw,bytecode benchmarking,genetic programming,performance prediction,reverse engineering},
mendeley-tags = {advert_pw},
number = {6},
pages = {865--877},
publisher = {IEEE},
title = {{Using genetic search for reverse engineering of parametric behavior models for performance prediction}},
url = {http://www.scopus.com/inward/record.url?eid=2-s2.0-78649781035{\&}partnerID=40{\&}md5=570c7f104b0d13913510278d9fff50ea},
volume = {36},
year = {2010}
}
Powered by bibtexbrowser