With a grid of machines available for benchmark evaluation distribution of workload is a clear way of optimization acceleration. Unfortunately numerous issues arise when extending MLO with this feature. When sampling the design space the active learning scheme has to take into account that there are design being evaluated - otherwise valuable resources could be wasted on redundant evaluations. Integration with non-uniform design spaces poses further problems..