I have another question.Is there any proper documentation how Flashphoner handles the servers that is attached to AWS load balancer?
Please see tthe section
Testing on
this page: "More instances will be started in case scaling is triggered. (For testing purposes, streaming with transcoding – e.g., streaming RTMP to
auto created mixer – can be used to load server CPU.) All the started instances will be auto added to the corresponding load balancer."
So if you created scaling group, you should not start new instances manually, AWS LB will start them as previous instances reach CPU load limit. New publishers and players will be routed to new instance.
1) Lets consider there are three servers A, B and C
2) A stream is published and it connects to server B
3) A viewer trying to view the stream and do we need to connect to the server B or connecting to Load balabcer endpoint will fetch the stream automatically from server B
If servers behind load balancer should be already running is is recommended to use CDN:
1) One LB is used for Origins and one LB for Edges
2) A publisher connects to LB websocket entry point, stream is published to some Origin
3) A viewer connects to another LB websocket entry point, stream is pulled by some Edge and played from it
Otherwise a player should be routed manually to the server where stream is published.
