[bestinterest]

Not sure I agree on the operational aspect in the large scheme of things. Servers are cheap, people are expensive.

Thinking out loud here, say you can only serve 200rqs with Rails but 2000rqs in some other framework (yes not fair comparison, endpoints matter etc etc). That means you need 10x the servers but take a step back a sec. 200rqs is 720k a hour or 17,280,00 a day or 518,400,000 requests a month. (The scale for that is pretty high already). I'm pretty sure the 10x increase you need in servers will cost less than 1 extra employee at say $70k a year.

[mohanmcgeek]

This has always been the argument of people optimising for development speed. Fair argument when you're actually getting dev speed in return. But it's mostly consulting companies making this argument, who don't have any skin in the game for ops cost.

If this were true, startups wouldn't be paying a third of the money they raised to AWS.. considerably more than people cost.

Companies grow and this 1/10th efficiency will quickly multiply.

[joelbluminator]

If ops cost are so important why are we all developing micro services now? That's 50 containers instead of one big one, extremely inefficient. Today I learned our 15 devs development environments cost as much as the whole production setup (!) because we each get a whole kubernetes cluster with all the containers. That's quite bonkers. So do we care about ops cost or do we not? I think we don't really.

[Nextgrid]

> startups wouldn't be paying a third of the money they raised to AWS

They're paying that because they and their investors want to. If you look beyond that it's absolutely possible to get that infrastructure at a tenth of the AWS cost by going with old-school bare-metal providers such as OVH or Hetzner, but then it means you no longer get to brag about how you wrangle (self-inflicted) complexity with Terraform and tons of YAML files at the next AWS conference.