Preparing infrastructure view...
Preparing infrastructure view...

Home / Infrastructure design / HPC
For research groups and engineering teams whose jobs do not fit on a single machine. Scheduler, fast interconnect and storage that keeps up.
What it is
For research groups and engineering teams whose jobs do not fit on a single machine. Scheduler, fast interconnect and storage that keeps up.
Slurm or Kubernetes, with queues, fair share and accounting.
Low-latency fabric sized to the communication pattern, not to a catalogue.
Throughput that keeps the nodes fed rather than starving them.
One cluster, partitioned so both kinds of job get a fair run.
How it works
In the field
What the work looks like once it is on the floor.


Detail
| Typical scale | From sixteen nodes, with head nodes and storage sized to match. |
|---|---|
| Scheduler | Slurm, or Kubernetes where the group already lives there. |
| Storage | Parallel file system plus a capacity tier for results. |
| Users | Accounts, quotas and accounting so usage can be reported per group. |
| Timeline | Twelve to twenty weeks depending on fabric and storage. |
Watch
Watch

Support
Fixed price against a defined scope. Ongoing operation is quoted separately as a monthly service.
We specify it and can procure it, but it stays on your balance sheet.
Yes. Design and build are delivered white-label where a partner holds the customer.
Next step
An hour with our engineers, no charge. Bring a floor plan, a workload, or just the problem.