Jupyter Experiences on Microsoft Azure

Lee Stott · ‎Apr 02 2020

The Jupyter Notebook is an open-source web application that allows you to create and share documents that contain live code, equations, visualizations and narrative text. Uses include: data cleaning and transformation, numerical simulation, statistical modeling, data visualization, machine learning, and much more

Requirements which we typically receives from education customers

•Multiuser Environment typically required

•1 VM server to class >100 student (Typically 1Gb to 2Gb of memory per student

•Offline or Online usage of Jupyter

•Container or scalable compute setup for larger classes <100 Students

•CPU and GPU requirements all required depending on DataScience or ML labs

•Custom Libraries and Environments are a given these are typically python PIP Install

•Persistent storage for storage of student's files and work.

• Understanding of cost and services utilization

So what are the options for solutions

1. JupyterLab

•Web-based interactive development environment for Jupyter notebooks, code, and data.

•Collaborative working environment for multiple users.

•JupyterLab is flexible: configure and arrange the user interface to support a wide range of workflows in data science, scientific computing, and machine learning.

•JupyterLab is extensible and modular: write plugins that add new components and integrate with existing ones.

2. Jupyter Notebooks

The Jupyter Notebook

•Single time notebook runtime environment can use various kernels Python, R, C#, F#, Q#

•Simple environment single view per notebook

•1 user interaction with the notebook

The two options for these within the classroom is typically in the form of IaaS based VM which can we be setup in a 1:1 or 1:many fashion for providing Jupyter experiences to students.

•IaaS based >High Admin Requirements most academics don’t want to manage infrastructure. However to enable the onboarding and setup of the environments we have produced the following templates and instructions in assisting you create a Jupyter experience for your classroom or student projects.

The Littlest Jupyter Hub
http://tljh.jupyter.org/en/latest/install/azure.html

Zero to Jupyter AKS
https://zero-to-jupyterhub.readthedocs.io/en/latest/

Azure Quick Start Templates
https://azure.microsoft.com/en-us/resources/templates/?term=Jupyter

However once you have your environment deployed there a number of tasks you or the IT team need to do in order for it to fully ready for your students.

Admin Tasks which need to be completed for classroom environment

Setting up Storage and sharing notebooks

•Distributing materials to users with nbgitpuller

•Adding data to the JupyterHub

•Share data with your users

Adding users to the Jupyterhub

•Add / Remove admin users

•Authenticate any user with a single shared password

•Authenticate using GitHub Usernames

•Let users choose a password when they first log in

•Let users sign up with a username and password

Configuring Environments