The Pos Experiment Controller: Reproducible & Portable Network Experiments

Network experiments are an essential tool to derive knowledge about network behavior. Yet, the experiment landscape is fractured, consisting of many specialized experiment enclaves, i.e., testbeds, with their own toolchain, limiting portability between different testbeds. Such fracturing hinders research, increasing the costs of replicating research across enclaves. This increase in cost also holds for the possible reuse of experiment parts across testbeds. Previous work introduced the pos framework and methodology, reducing the required effort for reproducible experiments. However, currently, this framework is available solely in a limited set of testbeds. In this work, we extend the proposed framework and, thus, build bridges between existing testbed islands. To showcase our approach and its feasibility, we recreate a well-known experiment on multiple testbeds and compare the results. These results indicate repeatability across testbeds. Concurrently, they highlight the flexibility gained through our approach and the need for reproducibility-aware experiment design.