With http live streaming (HLS) we need to deliver video segments one after other to the player to play them in a continuous fashion. Segments are present on many streaming servers and we would load balance between these servers to deliver segment.
Assuming that every request for chunk hits our application server and we redirect to best possible delivery node, once we get a segment request. Will it lead to opening of many TCP Sessions for the HTTP request handling. For example, there would be 1500, HTTP chunk request (TCP Sessions) per user-session while watching a 2 hrs movie. If so this can be in-efficient and is there a way such that we would redirect every chuck request to the delivery node and maintains same TCP Session?