juice, I get what you're saying, but even with variability, these numbers should at least be somewhat consistent across tests. If I run the same proxy twice, I'd expect similar results, no? The fact that I get a rollercoaster makes me think something's off, maybe server load or how the tool...