From reading through the forums it seems that one server seems to max out around 100 concurrent incoming streams. We are currently right around that number and slightly over at times and are trying to develop a plan to go to 200+ simultaneous incoming streams. Is it possible to have multiple origin servers where all broadcasters can connect to one rtmp URL? Maybe in a 2 origin and 1 edge configuration where all users connect their encoders to the same rtmp URL. Then, a load balance type system routes the incoming stream to one origin or the other?
We are trying to avoid the need for some users to connect to rtmp://origin1.domain.com and some users connect their encoders to rtmp://origin2.domain.com.
The Wowza load balancer is for distributing client connections to edge servers in a live repeater or MediaCache edge cluster. The live repeater system has a way to fail-over to a back up origin.
Thanks for the quick reply and helpful info. As a follow up question, would a server load balancing type of system work where we basically mirror the existing server and use some sort of server load balance system?
This is very helpful for our discussions here. One new question… If connecting to different origin addresses like rtmp://origin1.domain.com and rtmp://origin2.domain.com was not an issue, what are some of the most common methods that people use to grow past the 100 concurrent incoming stream maximum?
One more quick question… If our server’s memory and CPU resources max out at 100 concurrent incoming “live-record” streams, would we be able to have more if they were simply “live” incoming streams (i.e. not recording to hard drive)? Basically the question is, does it take more RAM and/or CPU to do “live-record” vs. just “live”? And if so, would rough percentage increase in concurrent incoming connections would we be able possibly see (5% more, 50% more?)?
Here is one more question to add to my post above… If RAM is our limiting factor for the maximum number of concurrent incoming streams, would we be able to get 50% more (or some percentage more) if all the incoming streams broadcast bit rate was lowered by 50%? Or, would is the incoming stream’s bit rate not a major factor in determining the maximum number of incoming streams a server can handle? I’m just trying to understand things a little as we make our plans to scale up.