Supporting NASA Science with OpenStack
summary
On‑prem OpenStack (Wallaby via RDO) runs ~300 hypervisors with about 15,500 cores and 75 TB RAM across three AZs, backing ICESat‑2 and the Nancy Grace Roman telescope with JupyterHub, Slurm, fast data‑transfer nodes and a bare‑metal GPU pool (88 V100s plus a DGX A100). Storage is NetApp AFF/ONTAP with NFS/Cinder, ~140 TB on flash and dedupe up to 20x; OpenID and end‑to‑end encryption are live. Operational pains: forced OS upgrades, RabbitMQ/telemetry instability, Linux‑bridge→OVN migration, xcat→ironic evaluation, and confusing flavor/AZ mappings.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 1:09Nancy Grace Roman Space Telescope
- 2:33Neutron Networks
- 4:17Ice Cloud Land Elevation
- 5:16HPC Center
- 7:20Base Deployment
- 10:44Cloud Overview
- 12:10Compute Flavors
- 13:11Networking
- 17:20Storage
- 20:48GPU
- 22:56Discussion
- 23:09The Good
- 27:58Other Challenges
- 31:15Moving Forward
- 33:39Closing
- 37:27BGP
related talks
16mA novel Cloud High Performance Computing infrastructure for user
36mOpenPOWER and GPU Porting Platform for FOSS Using OpenStack at OSUOSL
13mAn open cloud for researchers, staff and students
27mKeynote: OpenStack and the Software-Defined Supercomputer
57mSnakes on a cloud: the OpenStack project
15mThe CERN Cloud Infrastructure: 5 years in 15 minutes
15mAn open cloud for european researchers, staff and students
26mBuilding the European Weather Cloud with Open source software : An end to end journey.