Every workload came out of the move with more processor and memory

Capacity that was being spent on the old platform's own overhead now runs your services.

Three appliances that existed only to monitor the batteries were retired, and the management server the old platform required was deleted — that alone returned 944 GB of fast storage and took one machine from 73% full to 3%, which is what made it usable as a target at all.

The reclaimed capacity was not left idle. The main cluster's control-plane machines went from 30 to 50 processors (+67%) and from 90 to 120 GB of memory; its workers went up 40% and 56%. The secondary cluster's control plane was deliberately reduced, because it had been oversized at the expense of the cluster that actually serves you.