They seem to cheat with their "Mean Web Response Time". As you can see in the graph, at the time the App Server Availability is down to 0, the Response Time is also 0, where it should be ∞, or at least the timeout time. This would reflect the real mean response time, or am I wrong?<p>Edit: It was just one spike that went down to 0, directly after it there's a higher spike. (It didn't change the mean response time of 113ms, though)