Data Science: JupyterHub 2.0 introduces role-based access control

Share your love

Project Jupyter has released JupyterHub 2.0. JupyterHub is a server running in the cloud that makes the Jupyter notebooks, popular in data science and machine learning, accessible to user groups instead of individual users. The new release adds role-based access control (RBAC) and relies on JupyterLab for the standard user interface.

The use of scopes or roles, which are a named collection of scopes, should enable a finer gradation of access rights in JupyterHub 2.0 and thereby increase security. This means that every user or service no longer has to receive full admin rights in order to carry out a specific action with elevated rights, as these can be specifically assigned thanks to the role-based access control. Admittedly remains admin continue to exist, but “shouldn’t have to be an admin anymore”, as it says in the blog entry for the announcement.

Due to this innovation, an upgrade to JupyterHub 2.0 should require a longer downtime: All services and single-user servers must be shut down, then the update must be installed in the user and hub environment at the same time. The JupyterHub team advises that the database must be backed up before any upgrade.

With the standard user interface (UI), the new JupyterHub version relies on the web-based user interface JupyterLab. If desired, the classic notebook server can also be used again:

c.Spawner.environment = {
    "JUPYTERHUB_SINGLEUSER_APP": "notebook.notebookapp.NotebookApp",

There is also a third option, RetroLab, which can be accessed using c.Spawner.default_url = "/retro/" activate. This UI, also in the classic style, is based on the newer Jupyter server like JupyterLab.

Read Also   Energy transition index: Germany needs smart meters and bidirectional charging

JupyterHub first appeared in 2015. The multi-user server runs both in the cloud and on its own hardware and can provide groups of users with a preconfigured data science environment for Jupyter notebooks. It is configurable, scalable and open source. Possible areas of application are studies or research, in which users can use their own workspaces on shared resources that can be managed by system administrators.

All further information about the second major version provides the Jupyter blog.


Article Source

Share your love