This paper investigates the question of improving a performance model by exploiting all the information that is known about a system. The modeling perspective is to be opened up to encompass the entire available dataset, which could make performance models a useful adjunct in big data analysis of performance. The scope and content of the model would not be limited by preliminary decisions made by a performance expert, but would be determined by the data itself. For this purpose it may be necessary to create multiple sub-models in various formalisms (in order to capture different kinds of mechanisms in the system), and to associate them. We consider how a decision can be reached to add another sub-model, and how it can be associated with the existing model. The paper is exploratory only; a principal goal is to identify fruitful research questions in this area.

Additional Metadata
Keywords Hybrid modeling, Performance data, Performance models
Persistent URL dx.doi.org/10.1145/3053600.3053651
Conference 8th ACM/SPEC International Conference on Performance Engineering, ICPE 2017
Citation
Woodside, C.M. (2017). Open-perspective modeling of software systems. In ICPE 2017 - Companion of the 2017 ACM/SPEC International Conference on Performance Engineering (pp. 211–215). doi:10.1145/3053600.3053651