Azimuth - self service cloud platforms for reducing time to science
summary
Azimuth provides a self-service portal that lets researchers deploy curated OpenStack-backed platforms (JupyterHub, Slurm clusters, Kubernetes) using Ansible, Terraform and AWX with images that register services to a Zenith application proxy so no floating IPs are consumed; GPU, SR‑IOV and RDMA are exposed transparently and Lustre mounts appear inside pods. Cluster API creates k8s clusters with kubeadm, Prometheus/Grafana/Loki handle monitoring/logs, and an autoscaling Dask demo actually spawns nodes and speeds up work. Operators keep control via templates while users get Azimuth portal access and per-user Jupyter auth through a remote-user header.
AI-written summary · about
chapters (11)
- 0:00Intro
- 1:11StackHPC Company Overview
- 2:08Dynamic HPC
- 4:51Managing the complexity
- 8:02Bigger laptop
- 10:04Dedicated Sturm cluster
- 12:49Applications on Kubernetes
- 19:47Zenith application proxy
- 21:10Summary and future plans
- 23:00Join us!
- 27:42Dedicated Slurm cluster
related talks
30mOnyxia: A User-Centric Interface for Data Scientists in the Cloud Age
29mXpanse - An Open Services Cloud Project
37mBuilding a self-service kubernetes platform for python developers
35mDeployment and Management in the Age of Cloud Integrated Applications
34mEmpower Developer Experience
33mMulti-cloud Management: Stratos and Kubernetes
30mSelf service OpenShift cluster creation
33mThe Future of DevPortals