I think the core advantages of micro-services are:
1. Polyglot applications. There often is some library in a language for a use case, but there is also often the library in a specific language - the one that has a far larger community around it, very active development, considerable battle testing, a wealth of examples and guides and support available online. If you use this library in this language, you just kind of know it will work as best and easily as possible. The trail is well trodden. You now have a new problem of API contracts and versioning and such, but pretty much every language can at least HTTP and JSON reasonably at this point, so it's fairly tractable.
2. You already have code written in another language that pretty much works, either because of (1) or for historic reasons. If you go all-Elixir, you're going to need to re-write that hairy-ass black-box of a Go service that interfaces with Chinese SMS delivery. But it pretty much works, and if you can just chat with that over HTTP you can keep going. However, you probably want something to turn that thing back on every week-or-two when it inevitably eats it for whatever reason, and K8S gives you a really nice interface for this.
In general, I think - and this was an opinion I read on this forum at some point - that when building from scratch it's best to build a mono-service and then break small things off from that as the need arises, and Elixir would probably be at the top of my current list.
However, if you can run that mono-service in a micro-service environment - Kubernetes and containers - then you're well prepared for breaking pieces off or rolling new isolated components using the "right tool for the job".
I know there are difficulties around BEAM and K8S with both kind-of trying to do a lot of the same things, and that's something to definitely be aware of... people that have worked with it seem divided on where it falls between terrible and not really that bad of an idea (if you're paying attention and have some idea what you're doing).
And especially now, when there are the de facto standard libraries to do various tasks (like TF in Python) I think language lock-in can be a terrible idea.
>> “High-performance polyglot VM. GraalVM is a universal virtual machine for running applications written in JavaScript, Python 3, Ruby, R, JVM-based languages like Java, Scala, Kotlin, and LLVM-based languages such as C and C++.
