One of the recurring questions in the GCaP class last week was: How can we make web-application load tests more representative of real Internet traffic? The sticking point is that conventional load-test simulators like LoadRunner, JMeter, and httperf, ...