| Grant number: | 19/02144-6 |
| Support type: | Regular Research Grants |
| Duration: | June 01, 2019 - December 31, 2021 |
| Field of knowledge: | Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques |
| Principal researcher: | Leonardo Montecchi |
| Grantee: | Leonardo Montecchi |
| Home Institution: | Instituto de Computação (IC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil |
Abstract
Model-based evaluation plays a key role in dependability and performability evaluation of systems. Modeling allows the system to be analyzed at different levels of abstractions and in different conditions, without actually exercising the real system. Stochastic Petri Nets (SPN) and their extensions are a widely used methamatical formalism to evaluate non-functional properties across different domains. They haveseveral key advantages: they provide a convenient graphical notation, they support different abstraction levels, and they are well-suited for the representation of random events (e.g., component failures). However, their application for the performability analysis of modern systems is facing great challenges: systems scale is growing, they are becoming massively distributed, interconnected, and evolving. The complexity given by the high number of components, their interactions, and rapidly-changing system configurations represents a notable challenge for the specification of such models. In the software engineering discipline there are well-established ways to cope with complexity. Concepts like modularization, reuse, composition, inheritance, etc., are some of the basic tools of any software developer. Similar conceptsare instead less established for the construction for performability models. The objective of this project is to define a method to support modularity and reuse in the construction and management of performability models, primarily in models based on Stochastic Petri Nets. We plan to address this problem using a model-driven approach, that is, combining domain-specific languages and model transformations.The methodology proposed in this project will be evaluated through its application to concrete use cases, and the comparison with the traditional process for model specification. (AU)