Servers and Compute (ProLiant, Synergy, Apollo)
What this domain tests
Servers and Compute tests selection, sizing, and management across HPE ProLiant, Synergy, Apollo, and Cray. You need to match form factor, density, I/O, and management to the workload, and explain why one platform beats another without overselling.
Core rules
- ProLiant DL is rack, ML is tower, and BL is blade; Gen11 adds iLO 6 and Silicon Root of Trust.
- Synergy is composable infrastructure: frames, compute modules, interconnects, and storage modules managed by HPE OneView.
- Apollo covers HPC and AI density: Apollo 6500 for GPU-rich nodes, Apollo 2000 for dense compute.
- Cray scales to exascale-class HPC and AI; do not position it as a general-purpose rack server.
- iLO Advanced and OneView change the operational story; licensing is part of the design.
Worked example
A research team needs eight GPUs per node for model fine-tuning, plus fast local NVMe. Reason: Apollo 6500 provides GPU density and PCIe capacity that a standard DL380 cannot match at the same node count. Pair it with high-speed fabric and HPE Services for deployment. If the requirement were 200 general virtual machines, ProLiant DL380 with OneView would be the better fit; Synergy would only make sense if composability and rapid profile changes were explicit requirements.
Traps
Calling Synergy a blade system; ignoring iLO licensing; sizing GPU nodes by CPU count; assuming air cooling fits every Apollo configuration; proposing Cray for ordinary enterprise virtualization.