Use case: give access to medium-large datasets hosted on the mesocenter
Scenario: a researcher has some medium to large non sensitive data that she wants to access -- and grant access to other researchers at Paris Saclay -- through Jupyter for moderate calculations.
Bonus: anonymous access to visitors
For an example, see this issue
Tentative solution:
- host the data on the mesocenter
- mount the data in Jupyter servers provisioned by JupyterHub@Paris-Saclay
In principle, we have all the bricks to support this. The difficulty is to make it easy for users to access the data. Two approaches:
- User-side mounts with, e.g. S3+fuse; difficulty: supporting fuse within containers; also authentication, though this is mitigated by the fact that the data is not sensitive; it can be ok to make it public.
- Server side mounts: either there must be some API/UI for the end-user to choose which data she wants to access; or there must be some /data directory which is always mounted and give access to all publicly hosted data on the mesocenter.
Modification effectuée par Nicolas M. Thiéry