Yeah, I don't really understand why people are making a big deal about this. Two lambda functions passing data to step functions? Not exactly the best serverles

Yeah, I don't really understand why people are making a big deal about this. Two lambda functions passing data to step functions? Not exactly the best serverless architecture. And if you think about the problem of checking video streams, there is really no reason to scale different parts of it independently because you just need one process to watch the video and check for errors. Plus they still deployed the new solution on ECS, so it isn't really a monolith application that runs on a fixed number of servers. It's a real nothing story - the Prime Video team was just doing some engineering maintenance. Honestly the bigger story is that it was mischaracterized from the beginning, but I would imagine that is more a result of things coming together from multiple different systems and teams more than the way someone designed this stuff.