Reading of the input datasets could be tremendously slow. I think we should do it in parallel
Reading of the input datasets could be tremendously slow. I think we should do it in parallel