RFC: preauth benchmarking methodology

[Nathaniel McCallum]

On Fri, 2011-06-03 at 16:48 -0400, Dmitri Pal wrote:
> Probably we should test with multiple worker processes.
> So this is another dimension to throw into the result matrix.

Given that multiple processes are entirely independent, I'm not sure we
gain anything by this.  Especially since our goal is to measure the
effect of an asynchronous event loop that lives entirely within one
process.  If we add multiple processes, we are no longer measuring the
event loop but system performance.