How can I measure the load capacity of my video encoding server and plan for scaling?

Question

I have a server running on Amazon EC2 that acts as a video encoder (LAMP CentOS + FFMPEG). How can I load test it with video files and get an idea of how many encodings it can handle. I can't seem to find where FFMPEG keeps a log file or how to enable logging with FFMPEG can anyone help?

Some assumptions -

1) Only 1 video can be encoded at a time, I have set up a SQS queue to feed the processes to the server. 2) I can use Amazon Cloudwatch and Auto Scale to simply start another instance when my SQS queue gets to a certain size or performance on server slows - Amazon will take care of distributing traffic between multiple servers.

what are best practices for scaling video encoding servers, am I better off having lots of small instances or fewer larger instances?

thanks

Answer 1

video encoding is almost always cpu bottlenecked, so odds are amazon's high-cpu instances will be worth it for you. Similarly amazon's pricing has clear economies of scale so fewer more powerful instances are going to be more cost effective in general.

Push comes to shove though, none of us know your app so you gotta test yourself and do the math.