Question

Design the session orchestration layer for a cloud-gaming (pixel-streaming) service that runs the actual game on a GPU server in a datacenter, encodes the rendered frames in real time, and streams them to the player's device while sending the player's inputs back upstream — all within a <60ms input-to-photon budget so the game feels responsive. You must place a player on a GPU instance near them, capacity-plan an expensive GPU fleet for spiky demand, encode game video with minimal latency, and handle the round-trip control loop.