how to simulate load on multiple app servers when nginx balances load according to IP

Question

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 ?

Answer 1

Use multiple load generating engines or put multiple IP addresses on the one you have.