I am using FFMPEG to encode a video stream that is sent to a server at EC2 through ELB. I was only able to get about 2Mbps/15fps throughput to the EC2 instance.

Through process of elimination, I identified ELB as the bottleneck, they provisioned it at a higher throughput and now things are great (getting a solid 4Mbps/30fps)

How could I identify the bottleneck with a tool such as Wireshark? Is that possible? The process I did for finding it was kind of clunky and sometimes the ELB would auto scale up and fix the problem temporarily. I would expect if I looked at traffic with Wireshark I would see some ICMP responses or something from whichever device is the bottlneck, but I'm not really sure.


I would take a capture on the client and one on a backend server at the same time. If you analyze the client capture and the delays seem to be server side, but you analyze the server side capture and the delays seem to be on the client side, then it's probably something in the middle introducing delay.

To find delays in a pcap, check out a video I made that shows you how

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.