I have a video feed coming off a camera. Now I want to encode this video stream into multiple formats so that it can be played on different devices. I want to do the encoding on the cloud. I have a few questions on the implementation:

I understand that stronger the VM better will be the encoding speed. But instead, can I split the live encoding job across multiple small VMs so that I can spin up and shut down the VMs dynamically depending on the need (specifically want to know if it is feasible for for live video).

Also, once i create a particular stream ( say for iPhone ), how will it affect the performance depending on the number of iPhone users accessing the stream. Is there a need to create multiple streams to satisfy the demand of growing iPhone users?

To do a small POC, can I use FFMPEG for the job? or what is the best way to implement the POC?

There are a few services online that do the encoding job. http://zencoder.com/en/live-transcoding , http://www.livetranscoding.com/ It would be great it if I can know, how any of these transcoders work.

Any help is much appreciated.