Being able to do this at the programming language level is extremely powerful and creates an entirely different way of building applications. My go-to analogy is building a city (lots of individual, isolated, separate stack processes) instead of a big skyscraper (deep stack requests, concurrency difficult, error recovery manual).
You can build that city in a limited way with k8s, but there's way more overhead along the way, to the point of it not being enjoyable for me.
Yes because concurrency and fault tolerance are literally at the core of the the BEAM. you get all the above with zero configuration, all through convention and you only need to learn one set of tools.
When some new container technology comes along, or new worker queue platform etc. Erlang/the beam will be exactly the same.
One of the advantages of working with Erlang I found is that it's just as easy to scale down as it is to scale up.
How does one scale down from Kafka once you have vendor lock in with them?
Being able to do this at the programming language level is extremely powerful and creates an entirely different way of building applications. My go-to analogy is building a city (lots of individual, isolated, separate stack processes) instead of a big skyscraper (deep stack requests, concurrency difficult, error recovery manual).
You can build that city in a limited way with k8s, but there's way more overhead along the way, to the point of it not being enjoyable for me.