Dask: We had dedicated infrastructures for loading the cases. So these things would have a run by a run of each. They then were loaded on to another node, which isn't on the regulation clock as it were. And then we can start actually decimating this data and getting what we want out of it.