I haven't noticed the same trend or evolution of application tiers, perhaps we live in different echo chambers. Teams using microsevices need to evaluate whether it's still a good fit considering the inherent overhead it brings. Applying a bandaid solution on top of it, if it isn't a good fit, only makes the problem worse.
I think the term "microservice" is useless here. It doesn't matter if you run your backend logic in a monolith or in some complex microservice architecture, because both will depend on external runtime dependencies. Especially the smallest of startups will rely heavily on external APIs to connect their monolith (which is, in enterprise terms, probably a single microservice) to external services such as stripe, to some product analytics tool, to a CRM, to openAI etc. etc. pp.
I don't know anything about this post's solution, but if it delivers on the idea to not having to worry that much about failed calls to 3rd parties (or even my own database!), I'd like it a lot. Why would you call that a bandaid solution?
