Index of Pangeo JupyterHub Deployments

This thread collects a list of pangeo JupyterHub deployments. Each response linked below will share information on:

  1. Who can access the Hub
  2. How to sign up for access to the Hub
  3. The deployment configuration, if public
  4. Datasets co-located with the Hub’s compute, if relevant

Please add your own deployments as comments below!


6 Likes

The Planetary Computer maintains a pangeo-style JupyterHub deployment at Hub | Planetary Computer.

  1. Anyone can access the Hub
  2. Fill out an account request form. New requests are generally approved within a day, and you’ll sign in with a Microsoft Account.
  3. The deployment configuration uses terraform and Helm and is available on GitHub at GitHub - microsoft/planetary-computer-hub: Configuration for https://planetarycomputer.microsoft.com/compute. The container images build off pangeo-docker-images and are built at GitHub - microsoft/planetary-computer-containers: Container definitions for the Planetary Computer.
  4. The Hub is deployed in Microsoft Azure’s West Europe region, and so has efficient access to all the datasets in the Planetary Computer data catalog.

This Hub is primarily for exploration and demonstration. We recommend that production workloads deploy their own Hub (see our deployment guide) or use one of the other compute options on Azure.

1 Like

GEOAnalytics Canada maintains a pangeo-style jupyter hub deployment amongst other tools for large Earth Observation data analytics at-scale on public cloud infrastructures.

Access to GEOAnalytics Canada is limited to paying clients, and all assets/datasets are private. You can contact us for further discussions.

1 Like

Maybe add the list of running 2i2c managed JupyterHubs? List of running hubs — Infrastructure Guide

By the way, how do you define “Pangeo JupyterHub”? Do you just mean JupyterHub + Dask + colocated storage used for geosciences compute (with Xarray & Co), i.e. the the Pangeo software stack?

2 Likes

We’re currently deploying Pangeo-style Jupyterhub on EOSC. It has already been used for several workshops and training, for example FOSS4G, or Clivar. Not sure about the future, but it would be nice if we could open it to more people.

cc @annefou @tinaok @PhenoloBoy

  1. Access to the Hub is currently limited to people participating to selected workshops,
  2. But if anybody else is interested to test or contribute, please contact us. Access is given through EGI check-in (European identity federation),
  3. Deployment configuration is described here
  4. Currently no dataset is really colocated with the Hub, but the goal is also to ease access to all European publicly hosted datasets.

We also have a Pangeo Jupyterhub at CNES (french space agency). It is an HPC deployment.

  1. Only people working on a project related with CNES can have access to the Hub (and the associated HPC platform).
  2. So you cannot just sign up.
  3. The Hub is accessible at jupyterhub.cnes.fr, but the configuration is not public. It’s mainly a virtual machine with Jupyterhub and Nginx installed on it, the VM being connected to the HPC facility for jupyter notebook server submission through Jupyter batchspawner. Pangeo software is installed as HPC modules and conda environments.
  4. The platform is collocated with PEPS, so it has basically access to every Sentinel 1 and 2 products. Other earth observations datasets are partially accessible.