At AWS on my service we have lots of microservices (and I think we need them at our scale), but they also share a bunch of the same code. So one problem is reas

At AWS on my service we have lots of microservices (and I think we need them at our scale), but they also share a bunch of the same code. So one problem is reasoning about when your random java library gets merged into some component and it takes a while to wind its way through pipelines and then subtely breaks something unexpected. Take a lot of engineers to deal with this architecture and I certainly wouldn't recommend it for most folks. In fact, one of the slightly hilarious things is that we started out with microservices and end up combining them together because - guess what - when you are building systems level software you really need to make inline calls instead of network calls (without timeouts, retries, circuit breakers, and all the BS). As a manager I really miss monorepos, postgres, and all the other simple stack stuff that makes developers super productive and IMO the work much more fun.