Cluster planning & supply
TRcompute supports equipment planning and procurement for HPC projects. We review node specifications alongside interconnects, storage and site constraints to define the hardware supply scope.
For a new cluster, share the applications, workload sizes and expected number of concurrent jobs. For an expansion, include the current node, network and storage specifications so compatibility can be considered from the start.
We help organise these inputs into a hardware brief and source equipment against the agreed requirements. The work can cover CPU and GPU nodes, storage, switches and the components that connect them.
| Hardware layer | Requirements to review |
|---|---|
| Compute nodes | CPU architecture, memory per node, accelerator requirements, local storage and node count. |
| Interconnect | Application communication pattern, network interfaces, topology, switches, optics and cabling. |
| Storage | Working data, scratch space, checkpoint volume, shared access, throughput requirements and capacity growth. |
| Site and expansion | Rack layout, power and cooling capacity, management network, installation access and planned additions. |
The procurement proposal should identify equipment, delivery assumptions and any approved alternatives. It should also establish who installs and configures the cluster, including the scheduler, shared filesystem and monitoring.
Benchmarking, commissioning and operational support require a defined scope and responsible party. Include these requirements in the enquiry so they can be addressed before an order.