I use httperf from external server to stress test my app (on scalr/EC2). However nginx is configured to distribute load according to IP address of the source (so each app server will receive all the traffic from the same user).
Any idea how to work around this and measure load handling when the number of app servers behind nginx increases ?