That's basically just like ETL with staging tables, and it's how I've typically done this stuff.

And it really proves the point you are making here. The sheer amount of engineering, overhead, etc. for these systems and it's like... why? Where's the win here?

Are there benefits to this model? Yes. Absolutely. YES.

But the risks and costs are so high, that only a very small fraction of use cases justify the effort and expense and likelihood of doing it wrong.

In general, this is how I feel about most of the "modern way" of programming. It isn't "MODERN". Most of it's quite old, and the only thing "modern" about it is the use of HTTP (which... 30-ish years old at this point?). But because some folks called it the "modern" way of doing things, and talked about the rewards without talking about the risks, a lot of folks who don't know any better, and were promoted far above their experience level (and possibly their skill level as well) said "sure, my 15 person company with revenues of $2m a year definitely needs to write our app to use event driven architecture and microservices for our 23 calls per day to our single app". And 5 years later no one can make sense of the system or maintain it because they reproduced the Netflix "death star" architecture.

J.Ja