Roles, tiers, and opex
You can buy all the servers in the world, but somebody has to rack them, wire them, and fix them at 3am. Every outage that should take an hour but drags on for four is the bill for not having enough hands on deck. Hire wisely: seniors are scarce and expensive, juniors fix less per shift, and every salary is real, recurring opex that lands whether the month was busy or quiet.
Different people do different jobs. Technicians build faster, network engineers unlock advanced products, and your reliability engineers are the ones who shrink the time-to-fix when a server falls over. As you grow, support and capacity planners stop the small fires from becoming big ones. A lean team is cheaper but punishing; a deep bench costs more but turns disasters into footnotes.
Detailed explanation
Roles: DC Tech (build throughput), Network Eng (advanced products), SRE (MTTR), Security Eng, Support, Capacity Planner. Each role comes in four tiers — Junior 1x / Mid 2x / Senior 4x / Staff 8x capacity slots — so one Staff-tier hire is worth eight juniors' worth of slots in that discipline. Salary scales roughly 5k -> 40k/mo, and Senior+ candidates expect a signing bonus ($15k-$50k) on top.
SRE slots drive repair speed directly: mttr_multiplier = max(0.3, 1.0 - sre_slots x 0.05), so each SRE slot shaves 5% off mean-time-to-repair down to a 0.3x floor. That multiplier is what converts a host failure's blast radius into either a brief blip or an SLA-breaching outage — the math ties straight into open incidents and component failures.