I see gaps in waterfalls with tests in internet explorer I don’t understand:
If these gaps between DNS lookup and connection initialization or between connection/ssl-handshake and ttfb would appear in follow-up resource requests I would assume the browser performs DNS pre-resolution and/or TCP pre-connection. But in initial (HTML-)request?
I don’t understand these gaps.
Is that a bug in rendering of wpt waterfall and dns lookup or connection initialization last longer than bars in waterfall chart look like?
Or is it a problem in wpt or browser software on agent machines and browser is busy in these gaps? But there is nothing eye-catching in cpu utilization…
Best regards, Nils