JupyterHub

JupyterHub#

JupyterHub Containers : Kubernetes Cluster

The DesignSafe JupyterHub operates on a Kubernetes-managed cluster at the Texas Advanced Computing Center (TACC), offering an interactive, scalable development environment that supports scripting, job orchestration, and visualization — all from your browser.

What Is Kubernetes?#

Kubernetes is a platform for managing containerized applications across clusters of servers. In this case, it ensures that each user’s Jupyter session runs in an isolated container, scheduled efficiently on shared compute nodes.

Resource Allocation#

  • When you start a session, Kubernetes provisions a dedicated container with:

    • Up to 8 CPU cores

    • Up to 20 GB of RAM

  • These resources are exclusively allocated to your session — no other users can access your container’s assigned CPUs or memory.

  • The underlying physical node is shared across multiple containers, so I/O operations may experience minor contention under heavy load.

Best Use Cases#

  • Development and testing of OpenSees workflows

  • Running OpenSeesPy interactively

  • Pre-processing input files and visualizing simulation output

  • Automating job submissions to HPC using Tapis

When to Transition to HPC#

If your workload requires more memory, cores, or multi-node execution (e.g., OpenSeesMP), you should move to TACC’s HPC systems like Stampede3, submitting jobs via:

  • Tapis from Jupyter

  • Tapis Apps via the Web Portal

  • Manual sbatch scripts over SSH

The JupyterHub environment offers an interactive and versatile platform that seamlessly bridges OpenSees development with Tapis-based HPC execution. Its tight integration and flexibility make it an ideal space for prototyping, testing, and scaling analyses — from initial development to full production workflows.

Contents