[MattRogish]

I agree - microservices are a technical solution to a people problem. Stevey's seminal "Platforms Rant" (https://gist.github.com/chitchcock/1281611) touches on this - the reason why Dread Pirate Bezos mandated services was not because Kubernetes was cool[1], but because their teams were too interconnected, moving too slowly, and Conway's law was getting in the way.

Splitting into services siloed each team and allowed them to move independently and by side effect, faster. Not due to inherent properties of (micro) services but because one goes faster by removing the things that slow you down.

As a startup, you do not have this problem. You likely will _never_ have this problem as your default future state is 'dead'.

Do the simplest thing that can possibly work - build a monolith using tools you know that give you the ability to predictably, rapidly iterate on the product.

After you hit some semblance of product/market fit and need to scale, you can do that. Scaling is a solved problem. Premature scaling is not.

[1. This is a joke. Kubernetes wasn't even a thought in Google's eye at this point in history]*

[rewmie]

> the reason why Dread Pirate Bezos mandated services was not because Kubernetes was cool[1], but because their teams were too interconnected, moving too slowly, and Conway's law was getting in the way.

Not quite. Amazon's problem was that they were experiencing too much impedance between teams, and some teams were even siloing themselves to the extent they were creating problems for everyone around them. Bezos' diktat was intended to break through a bunch of petty office politics bullshit that was dragging down the company and preventing teams from delivering their work. The diktat boiled down to basically ordering everyone to grant access to the service they owned and provided to external teams that need it, no exception, and would be held liable if they failed to provide it, intentionally or not