Hi there,
I want to deploy my Wowza servers on AWS EC2 instances. I’m calculating how many instances I need. On the Wowza manager website I see some indicators like “Wowza CPU, total memory…”. So for one application with one incoming rtmp stream (one programme), when the transcoder is enabled, how much memory and CPU resource does this application consume? It’ll be better if someone could give the approximate maximum application number on an AWS m3.large instance.
Further more, how to estimate the maximum number of concurrent .ts request one wowza application support? I have to decide the number of instances not only on published streams but also on client number.
Thanks! 
Yorick
Hi Yorick,
This would really depend on the video quality of your source stream and the number/level of encodes you plan to output. I would suggest spin up an instance, transcode your desired output, and run the Load Test tool. Periodically review your access logs found in [install-dir]/logs folder. If you find SKIP1FRAME messages, this would indicate that your Transcoder is struggling, and you should dial your output renditions back down, until you reach that optimal level of maximum transcoding quality for the given infrastructure.
Michelle
Hi,
The Skip Frame messages will be based on the transcoder load only (input and output) and not dependent be effected by how many clients are concurrently connected to the server.
The higher the bitrate, resolution and complexity of the source stream, the more CPU is required to decode the stream. Equally the higher the encode bitrate, resolution and complexity the more CPU is used.
For bandwidth you can refer to the WowzaStreamingEngineforEC2_UsersGuide when searching for “Performance benchmarks”.
Regards,
Jason