An industrial case study of performance and cost design space exploration

Abstract

Determining the trade-off between performance and costs of a distributed software system is important as it enables fulfilling performance requirements in a cost-efficient way. The large amount of design alternatives for such systems often leads software architects to select a suboptimal solution, which may either waste resources or cannot cope with future workloads. Recently, several approaches have appeared to assist software architects with this design task. In this paper, we present a case study applying one of these approaches, i.e. PerOpteryx, to explore the design space of an existing industrial distributed software system from ABB. To facilitate the design exploration, we created a highly detailed performance and cost model, which was instrumental in determining a cost-efficient architecture solution using an evolutionary algorithm. The case study demonstrates the capabilities of various modern performance modeling tools and a design space exploration tool in an industrial setting, provides lessons learned, and helps other software architects in solving similar problems.

Performance assessment of an architecture with adaptative interfaces for people with special needs

“…They could implement the system and the models accurately reflected the real system performance. -The work of de Gooijer et al (2012) re-architects a legacy system, for remotely diagnose industrial devices, in ABB company. The goal was to improve system performance and scalability.…”

Towards software performance engineering for multicore and manycore systems

“…Their models can be used in multi-objective optimization to find an optimal tradeoff between performance and costs. de Gooijer [50] et al used evolutionary algorithms to determine the optimal number of cores with respect to performance and costs for an industrial web-based architecture.…”

“…In general, it is still expensive to build predictive performance models [50], and this is even more true for multicore models, which are more complicated than classical single core performance models. Besides the need for welleducated and experienced performance experts, who are able to find suitable abstractions and have a feeling for the important performance issues in a system, it is also necessary to reduce the overheads for manually creating models based on measurements.…”

Rapid performance modeling by transforming use case maps to palladio component models

“…It offers the comparison and selection of several design alternatives of a software system already in the modeling state. The PCM has recently been applied in a number of industrial case studies [14,8,23].…”