Towards Scalable Search-Based Model Engineering with MDEOptimiser Scale
Alexandru Burdusel, Steffen Zschaler · 2019
Running scientific experiments using search-based model engineering (SBME) tools is a complex task, that poses a number of challenges, starting from defining an experiment workflow, to parameter tuning, finding optimal computational resources to run on, collecting and interpreting metrics and making the entire process easily reproducible. Despite the proliferation of easily accessible hardware, as a result of the increased availability of infrastructure-as-a-service providers, many SBME tools are rarely using this technology for accelerating experimentation. Running many experiments on a single machine implies much longer waiting times and reduces the ability to increase the speed of iterations when doing SBME research, thus, slowing down the entire process. In this paper, we introduce a domain-specific language (DSL) and a framework that can be used to configure and run experiments at scale, on cloud infrastructure, in a reproducible way. We will describe our DSL and framework architecture along with an example to showcase how a case study can be evaluated using two different model optimisation tools.