> Am 21.06.2021 um 18:25 schrieb Shawn Heisey <hapr...@elyograg.org>:
>
> On 2021-06-20 06:03, Shawn Heisey wrote:
>> Unrelated, and off topic because it's mostly about Apache, but strange:
>> I've been doing some tests with webpagetest.org, and seeing REALLY
>> long load times for some resources in their waterfall graph. I see no
>> speed problems when I load the pages from my workstation at home.
>
> Followup on this, information which others here might find useful:
>
> By default WebPageTest defaults to traffic shaping of 5 Mbps down and 1 Mbps
> up, which it thinks simulates a cable connection. That's laughable -- I get
> 460 Mbps down and 12 Mpbs up on my cable connection, and I'm not even paying
> for the maximum bandwidth I COULD get.
>
> Long story short, hitting a web page with about 25 megabytes of images takes
> over 40 seconds for WebPageTest to render. If I switch from that default
> "5/1 Mbps Cable" traffic shaping to native (no traffic shaping at all) the
> render takes 1.8 seconds, which is approximately what I see when I hit the
> page myself. Server in AWS.
>
> When I do the math, 40 seconds is actually quite fast for downloading those
> images on a 5 megabit connection. So there was no actual problem. WBT needs
> to make the choice of traffic shaping a lot more prominent, and provide more
> realistic options than what they have at the moment. To even see bandwidth
> options, you have to open advanced settings. And the only option I could see
> in their list that's faster than the default (aside from native) is FIOS,
> which they've got at 20Mb down and 5Mb up. They have forums, I'll make
> suggestions there.
>
> Thanks,
> Shawn
>
It’s probably to make DDoSes more difficult (like basically everything these
days)
I never got around to host my own WPT instance (for work).
I mainly use the public version to to get „a feeling“ for the speed and to weed
out any caching effects of local browsers with pages too complex to use curl or
httpie….
