Decent write-up. I agree that most things should start as a monolith and only move to microservices if no other solution will solve the problem. You can scale a monolith quite effectively with a little planning.
Agreed, but I have seen numerous times (I contract and see a lot of shops inhouse) that employees suggest microservices as architecture from the start for new projects. I have a strong feeling this is CV driven development, a.k.a "its trendy and I need skills in this field to stay relevant on the job market".
I joined a shop where that was the case. We invested time in reintegrating services back into the monolith. A number of them could be replaced with ~100 lines of ruby.
