Infrastructure for reproducible data analysis

Whole Tale, Jupyterhub, Binderhub, and other tools are important for reproducible data analysis. Currently they have to be hosted and run in a cloud somewhere; there’s no canonical hosted service available. As a result, organizations that want these sorts of services are getting sucked into multi-year contracts with big proprietary players who bundle such things with existing large-ticket services.

How can we make it easy for every academic institution to run their own instance or support a free-knowledge network sercice? What options and services and commitments are needed, where to start, who is doing this now?

See Chris Holdgraf’s comment here: