Choosing a cloud stack for a scaleup

Every stack comparison ends up as a features table. Ours usually starts somewhere else: who's actually going to get paged when something breaks, and how much of the stack can that person realistically hold in their head?

Start from the on-call reality

A small team benefits from fewer moving parts, even at the cost of some flexibility. A managed platform that trades configurability for fewer 2am surprises is often the right call for a five-person engineering team, even if it wouldn't be the right call at fifty.

Match the stack to the team you have, not the one you plan to hire

It's tempting to build for the team you'll have in two years. In practice, the infrastructure you can operate well today is worth more than the infrastructure that's theoretically ready for scale you haven't reached yet.

What tends to matter more than the provider

• How much of the setup is documented somewhere other than one person's memory.
• Whether a deploy can be reverted in minutes, not hours.
• How much of the monitoring is actionable versus just noisy.

Get those three right and the choice of provider matters a lot less than most comparison posts suggest.