GTC 2020: Improve Data Scientist and IT Productivity with Jupyter-as-a-Service
After clicking “Watch Now” you will be prompted to login or join.
Click “Watch Now” to login or join the NVIDIA Developer Program.
Improve Data Scientist and IT Productivity with Jupyter-as-a-Service
Emily Watkins, Pure Storage
In this talk, you will learn how to combine JupyterHub with the Pure Storage Orchestrator (PSO) to deploy Jupyter-as-a-Service. JupyterHub provides a single, central notebook platform that can be easily secured and managed. PSO is used to automatically provision persistent volumes on Pure Storage FlashBlade on-demand for new users. To further support these exploratory developer workflows, you can also integrate the Rapid File Toolkit that allows data scientists to scan through high volumes of data with a 50x speed compared to traditional Unix commands. We will examine how this “Jupyter-as-a-Service” deployment inside an AI Data Hub enables users to access both required python libraries and the organization's shared datasets from their persistent home directories. We will demonstrate how this deployment provides better CPU/GPU and memory management along with a persistent data platform for the entire machine learning pipeline.