In the waterfalls, we’re noticing some gaps between the initial connection (orange bar) and time to first byte (green bar) in which nothing seems to be doing anything. An example can be seen here: http://www.webpagetest.org/result/121015_BQ_C39/5/details/
We have older tests which show no gaps, e.g. here: http://www.webpagetest.org/result/120927_57_EDD/1/details/
We thought it might be something to do with CPU utilisation, but it doesn’t seem maxed out during these gaps (mainly lines 4-8).
Is there an explanation behind these gaps, and/or any way we can reduce or remove them? They seem as if they might be having an impact on our overall page loading times.
Also talking about CPU usage, to save creating another thread, this set of tests has full CPU usage right from the start to the end of page load: http://www.webpagetest.org/result/121016_63_81H/ - this doesn’t look right to us; is the CPU actually maxed out, or perhaps we’ve done something wrong with our test?