As I read more and more regarding actor pattern in stateful service, I am not convinced why Actor pattern is good solution with stateful services.
Most of the use cases talk about single-threaded, state managment of actors with short duration of operation in each actor, best scenarios would be small sized calculation in each actor with big numbers in actors operating in parallel.
In such scenarios, the bottle neck would be write operations from lets say millions of actors to the DB. The argument would be storing the states or small sized outputs from each actor to its own state manager or reliable collections such as reliable dictionaries, but still that is different from true stateful service using database.
I think this confuses use case recommendation of stateful services relying on DB vs. true actor patterns.
It would be great if someone can help me understanding what would be good use case of reliable actors with database in stateful service or please correct me if I am not understanding some aspects of reliable actors correct.