3 ms·
> I do think microservices were over sold as the answer to everything, and I think this may have arisen from vendors who wanted to sell Kubernetes with a simpl
by EMM_386 3y ago
> I do think microservices were over sold as the answer to everything, and I think this may have arisen from vendors who wanted to sell Kubernetes with a simple marketing message that enterprises needed to modernize by using Kubernetes to do cloud native microservices for everything
Yes, they absolutely were oversold, Kubernetes marketing was part of it. There was also a lot of resume-driven development, and simply "look, new shiny object!". A lot of young engineers entered the workforce over the last couple of decades. That often comes along with "forget this old stuff, what's the new stuff?".
I use KISS as much as possible with every project I'm involved in desinging.
The microservices push got the point where a dev shops with 10 people had it all broken into microservices and running Kubernetes to manage low-traffic front ends, an API and a database. I would see some of these systems and the only thing I could think was "... but why?".
Do microservices have a place? Yes. Have I seen them used where they shouldn't be? Yes. Have I seen more poorly implemented microservices architecture than good ones. Yes.
Over-engineering the tech stack, or selecting one part of it vs. another based on which is trendier, only leads to more frusturation, difficulty reasoning about it, added development time ... all of which means it will cost more.
- t0b1 3y agoThat reminds me of a somewhat funny story. I was listening to a presentation about a startup with about 10 people as you mentioned doing hospital digitalization (gist of it seemed to be sending patient data from one doctor to another). They explained that they wanted to choose a “robust” architecture so they chose microservices (kubernetes and all). I thought that was a bit odd for something hosted at a hospital locally especially because they had like 7 or so services (and maybe even two databases but not sure on that one). Well, we later asked them how much data they were even handling and (after a bit of side-stepping) they said it was around 200mb excluding images. That was my “but why?” moment. Apparently you need 7 services to have a web frontend for 200mb of data with most of it probably never accessed and hospitals can suddenly get three new floors overnight.