Eighteen services per developer, twelve developers, one box
Your platform is eighteen services. Locally, a developer runs four of them and mocks the rest, because eighteen containers on a laptop is fan noise and a swap file. A shared dev cluster would fix it, but a per-node bill for twelve developers' idle stacks never made it through budget.
On Hoody every developer gets the whole system: eighteen containers copied from the team's template snapshot, one set per engineer, two hundred and sixteen on a bare metal box. They share almost all their disk blocks copy-on-write; the kernel caches the common binaries once; a service nobody is calling burns no CPU. Each engineer's API talks to their own queue, their own worker, their own database, at their own URLs.
"Works on my machine" stops being an excuse when everyone's machine is the real one.
The dev environment used to be a subset; now it's the system.