for your last point, that the api is simple, explains all the errors, but forces you to handle them; that’s the aws philosophy. they give you full insight into how each of their products fail and it’s up to the integrator to figure out how to use them properly in a partial outage. it’s difficult to do this because it implies you’re code has to be multi region ready. contrast this with google that mostly designs the api as one that will try to mask as much of this as possible. they are trying to offer global solutions that solve it for you