Docker container vs JupyterHub

Apologize for the naïve question. I am setting up a 2 user data-science platform (as a POC for a larger application) . From my application (MEAN stack) on AWS EC2 I launch the JupyterLab in docker container corresponding to the user .

User 1 (Data Scientist) → Logged into MEAN Stack application → Choose image ( jupyter/datascience-notebook) → Clicks “Launch Workspace” → Launch JupyterLab (PWD /home/user1)

User 2 (Data Engg) → Logged into MEAN Stack application → Choose image ( jupyter/r-notebook) Clicks–> “Launch Workspace” → Launch JupyterLab (PWD /home/user2)

Can I achieve this using JupyterHub.

I installed JupyterHub (using Install JupyterHub and JupyterLab from the ground up) and created 2 users.

Would this JupyterHub + 2 users work as above as in 2 docker containers running separate images.

Do I have to install conda packages for these 2 users separately or is it possible to launch a chosen image in JupyterHub.

I have minimal AWS EC2 instance (not enough for Zero to JupyterHub with Kubernetes).


JupyterHub has lots of pluggable options, including for authentication and spawning servers. It sounds like you’re interested in DockerSpawner.