Has Pangeo gotten NSF feedback to use pre-existing infrastructure?

hey all - someone in the Jupyter community meeting just made a point that I think relates to Pangeo, and I’m curious what this community has experienced:

One of the participants mentioned a hosted / cloud-based analytics platform for molecular modeling. It required scalable computing and large datasets, and they wanted to provide Jupyter environments to let users interact with these.

They got a lot of generally positive response, and wrote up an NSF proposal to build a Pangeo-like model (Jupyter/BinderHub etc) for shared infrastructure within the project.

However, they didn’t get the proposal funded, and a big part of the response was “you should use pre-existing NSF infrastructure for compute etc, not use JupyterHub or cloud computing”. In their case, this means they now integrate with CyVerse, which is a big NSF-funded infra project (though doesn’t use many open source infrastructure tools and doesn’t use enterprise cloud at all, I believe)

I am curious if Pangeo has ever gotten feedback along these lines - and if so, how you responded to it?

1 Like

We have not yet proposed to NSF to operate infrastructure at scale for a large community. We would probably get such questions if we did. Most of what NSF is funding us to do is to develop tools, not operate them.

In a sense, we do operate with existing NSF-infrastructure–one of our biggest “deployments” is on the Cheyenne supercomputer.

NSF does not yet have a coherent strategy for how to support infrastructure on commercial cloud. @jhamman and I are attending this workshop next week, where these questions might be discussed.

I want to figure out how to keep the cloud infrastructure going and expand its user base, because I really believe it is the way forward. So I am very interested in this topic.