Tests the scalability of the data loading, running an MRP plan (including the clustering algorithm) and saving the plan. The network in this case consists of a lot of parallel clusters, which can be solved in parallel. See also the test scalability_2
The algorithms scale linearly with the model size, while the mayor underlying data structures are binary trees which scale logarithmically with the model size… The result is a runtime that combines both. In summary, one could say that the system scales a bit worse than linear, but definitely not quadratic or worse.