Create a local copy of the file as it is being dowloaded. Then, the next time the dataset is being requested, a local copy exists and can be used. Having files exclusively exist remotely only makes sense with a fast internet occasion, and with limited local storage capacity.