4 comments

[ 3.0 ms ] story [ 16.3 ms ] thread
I'm quite excited about finally being able to achieve the holy grail of "reusable microservices".

I think the key thing that makes this actually possible in the Kubernetes world is the idea of operators + CRDs. It gives devs/ops a stable surface for configuring/scaling/operating a reusable component which is a pain otherwise.

Perhaps even, the ease of running multiple microservices on Kubernetes will one day be greater than the ease of just operating a monolith.

Some reason I don't think free web and google should be in same clause.

They are like any other corporation who steal, spy, sell your privacy data and restrict you to use they silly sandboxes

Assuming the microservices are stateless, this is basically dynamic functional programming. It's how I like to program after spending some time with Common Lisp.

It does pain me to think about the massive overhead of communicating via HTTP, though. Just think of the whole network stack that is being used when doing one simple call and compare that to calling a subroutine on the same computer. It's really nice for us programmers but it strikes me as being too easy. Imagine being a civil engineer but you can make everything out of titanium or any other expensive material.

Modern HTTP has been made fairly efficient, especially with HTTP/2. I think the overhead is more in the data formats that are in common use, like XML/JSON, but with adoption of GRPC which combines Protobuf with HTTP/2 I think a lot of this is going to be more efficient. It's not as efficient as it could be, but we're also getting a lot of features and functionality for it too.