Automated Planning for Software Architecture Evolution
Jeffrey M. Barnes, Ashutosh Pandey, and David Garlan
In Proceedings of the IEEE/ACM International Conference on Automated Software Engineering (ASE’13), pp. 213–223
Abstract: In previous research, we have developed a theoretical framework to help software architects make better decisions when planning software evolution. Our approach is based on representation and analysis of candidate evolution paths—sequences of transitional architectures leading from the current system to a desired target architecture. One problem with this kind of approach is that it imposes a heavy burden on the software architect, who must explicitly define and model these candidate paths. In this paper, we show how automated planning techniques can be used to support automatic generation of evolution paths, relieving this burden on the architect. We illustrate our approach by applying it to a data migration scenario, showing how this architecture evolution problem can be translated into a planning problem and solved using existing automated planning tools.