Dedicated Jupyter Servers

The default Jupyter environment available to all Chameleon users is a bit limited: you are working within a shared environment and as such there are some practical limitations around the amount of CPU cores and memory you can utilize. More intensive analytical workflows may function better from within a dedicated Jupyter server for use by you and/or other members of your project.

Using the Appliance Catalog

The Chameleon Appliance Catalog provides a JupyterHub appliance that is functionally equivalent to the shared Jupyter environment. The appliance will allow you to reserve a Chameleon bare metal node and provision it with the JupyterHub application, along with a Floating IP Address that allows access over the public Internet. Any Jupyter Notebook servers managed by this multi-user environment will have access to the underlying resources on whatever node you have reserved, removing the limits around CPU and memory usage.

Using the Sharing Portal

The Sharing Portal also has a JupyterHub artifact you can instantiate on Chameleon. There is no material difference between this method and the Appliance Catalog method and it can serve as a nice introduction to the Sharing Portal if you are not already familiar with it. With this method, you actually provision your own JupyterHub server via the shared default Chameleon JupyterHub; it’s JupyterHub all the way down!