Introducing CIRRUS: NSF NCAR’s new on-prem cloud service

CISL recently launched the CIRRUS platform, cloud infrastructure for deploying research services by the NSF NCAR and university communities. This new on-prem cloud service is designed to seamlessly integrate with NSF NCAR's existing high-performance computing (HPC) infrastructure, support software development, and host various research applications for Earth system science.
 

“GDEX showcases the CIRRUS platform as the next-wave new-technology stack developers can leverage to provide capabilities that complement existing HPC infrastructure.” 

— Nick Cote, CIRRUS Technical Product Owner

CIRRUS is a Kubernetes-based, developer-friendly technology stack. The first primary use case for CIRRUS is the new flagship Geoscience Data Exchange (GDEX) data commons, which officially launched on September 9, 2025 (see related CISL News story). 
 

According to Varsha Banda, HPC Systems Engineer, the on-premises cloud foundation for CIRRUS is provided by clusters built on dedicated bare-metal servers. These servers are hosted at both the Mesa Lab and the NWSC.
 

“CIRRUS is both hardware and software infrastructure.” 

— Varsha Banda, HPC Systems Engineer

Each site runs a Kubernetes environment on physical hardware with high speed-connectivity to NSF NCAR’s shared Ceph storage and high-performance computing systems, Banda indicated.

The CIRRUS platform allows GDEX to deliver faster and more reliable service, and provides a platform for future AI features, said Nick Cote, the platform’s Technical Product Owner: “CIRRUS opens up so many different, flexible, interesting ways that we can continue to improve the GDEX platform as we move forward.”
 

“This cloud-native way decreases the time to delivery. It allows us to be more flexible in reverting changes, having version control, and how the codebase is rolled out, and it increases speed as well.” 

— Nick Cote

Cote continued: “If you're looking to host a website that does interactive web visualizations for a dataset that ties directly into your science, the CIRRUS platform can provide that directly out of your code repositories, which wasn't available before.”
 

Another advantage, according to Kevin Hrpcek, HPC Systems Engineer: CIRRUS provides a platform that promotes the automation of development workflows that includes code testing and dedicated development instances with minimal overhead. 
 

“Once automated, entire environments can be created or destroyed within minutes.”  

— Kevin Hrpcek, HPC Systems Engineer

CIRRUS, which supports continuous integration and continuous delivery (CI/CD) workflows through GitHub actions, is also being used for smaller projects: “We can host self-hosted runners through this really cool cloud-native mechanism,” said Cote. “We've got probably a dozen different NSF NCAR repositories connected to CIRRUS for that particular aspect, and probably just over 20 different user applications hosted as well. And part of those 20-plus involve different components of the GDEX project.” 
 

Platforms to leverage the CIRRUS infrastructure include the SAGE Data Commons and the modernized Research Data Archive (both now part of GDEX), various JupyterHubs (including one for classroom use), and custom interactive web visualizations. For example, SkyWash—Air Quality and Sky-Seeding runs on CIRRUS, offering real-time PM₂.₅ pollution tracking and washout simulation.
 

Cote shared what inspires him most about the CIRRUS platform—that is, the speed and efficiency advantages it provides to developers: “It's a much more developer-friendly technology stack. I have really been excited to get the developers using the CIRRUS platform, then seeing how the work that we're doing accelerates as people become more familiar with the new technology stack. It allows them to focus more on the code base and less on the implementation and configuration of a lot of different things. I think what takes up the majority of developers’ time at work, potentially 75 percent, is updating things that the CIRRUS platform eliminates.”

The team plans to continuously add new features in the future to streamline the user process and explore ways to better integrate CIRRUS with NSF NCAR’s HPC resources, according to Hrpcek.

The CIRRUS platform is currently available, added Hrpcek, and the team is actively promoting it to encourage the migration of applications and workflows. This new resource is open to all NSF NCAR departments and collaborating universities. 

Developers and researchers interested in the platform should contact the CIRRUS team.