Forrest Burt opens the main Fuzzball interface and contrasts it with the terminal-centric Linux environment HPC users know. Everything in Fuzzball is API-driven: the graphical interface wraps API calls behind buttons and switches, the command line issues the same requests, and a Python SDK lets teams submit workflows programmatically.
The practical upshot is that users never SSH into a cluster or manage a home directory. They work in a dedicated computing environment where they can attach volumes, pull images, and use resource definitions their administrator has added, including different GPUs when a workflow calls for them. The clip comes from a session setting up an AI workflow.
Key takeaways
- Every action in the Fuzzball GUI is an API call, and the same requests can be sent from the CLI or a Python SDK.
- Fuzzball users never need to SSH into a cluster or deal with a home directory.
- Within the interface users attach volumes and pull images, while administrators add resource definitions such as new GPU types.
Questions this video answers
Do Fuzzball users need to SSH into the cluster?
No. Fuzzball is fully API-driven, so the web interface, the command line, and a Python SDK all send the same requests to an endpoint. Users work in a dedicated environment where they attach volumes, pull images, and pick administrator-defined resources, with no SSH sessions or home directories involved.
This video is part of the Fuzzball playlist. Browse every CIQ video by product and topic.
Transcript
so yeah this is what the main interface of fuzzball looks like as said as opposed to the standard kind of Linux graphical terminal environment that we're used to in HBC everything in fuzzball is apid driven so we take API calls versus an endpoint we can wrap those and essentially anything we want in this case it's a graphical user interface with buttons and knobs and switches that we can push and Tinker with in order to send those API requests out if we were on the command line we'd be typing commands we also have a python SDK that you can utilize to programmatically submit workflows but point
being in the end you never have to be sshing into a cluster you never have to be dealing with a home directory anything like that you are in a dedicated Computing environment where you can attach volumes pull images your administrator can add new resource definitions if you need different gpus that type of thing um so just to kind of show you guys what fully is going on here
Built for scale. Chosen by the world’s best.
2.75M+
Rocky Linux instances
Being used world wide
90%
Of fortune 100 companies
Use CIQ supported technologies
250k
Avg. monthly downloads
Rocky Linux
9
Enterprise products
Spanning the kernel to the orchestrator
Have questions about your infrastructure?
Talk to a CIQ engineer about Rocky Linux, HPC, and AI infrastructure.