IANAS either, but mean values hide a lot of very interesting data.
Have you got any raw data going spare that I could Splunk?
On 22/11/2011, at 17:23, Mauricio Freitas
Michael,
Thanks for your questions. I'm not collecting this data myself, it's part of Google Analytics. It was by invite only until a couple of days ago. I've been using for a while, so had a good number of samples already.
IIRC the data is collected when OnLoad event is fired. Yes, I have all that data and could cross if you want to.
The report is about User Experience, not about line speed. "What's the average user experience, measured in time to load a page, in such a city/ISP" is the question to be asked here.
The question if the ISP A or ISP B attracts an audience with more or less updated machine is not part of the thing. I'd like to believe all ISPs would have a representation of our society in general. I'm no statistician and I look at the number that tell me on average a user on Orcon will see our page take almost 12 seconds to load, while a user on nzwireless (with a close sample size) can have that in 2 seconds.
As for your worry about Orcon (which seems absurdly high)... We have an interesting audience, as in most of our readers have a different browser profile than Trade Me for example. We currently have a mix of Firefox (29.74%), IE (28.43%) , Chrome (27.36%), Safari (10.33%).
More specifically in Orcon's case, their customers use Windows (68.70%), Mac OS (21.33%). In terms of browsers they are split in Firefox (31.41%), Chrome (30.36%), Safari (17.82%) and IE (16.73%). I don't think they have slower PCs than the average New Zealander. Their customers are actually savvy enough to move away from IE and even have a good Mac OS uptake.
Remember this is web page load time, which includes the html request, followed by all resources including images, scripts, CSS, ads. Some of these resources are local, some are from third party. Those resources however are presented as the same to everyone.
Cheers
Mauricio Freitas www.geekzone.co.nz www.freitasm.com www.twitter.com/freitasm
-----Original Message----- From: nznog-bounces(a)list.waikato.ac.nz [mailto:nznog-bounces(a)list.waikato.ac.nz] On Behalf Of Michael Fincham Sent: Tuesday, 22 November 2011 12:36 p.m. To: nznog(a)list.waikato.ac.nz Subject: Re: [nznog] Proxy Servers...
On Mon, 21 Nov 2011 15:18:14 -0800, Mauricio Freitas wrote:
Web page load time *IS NOT* time to download HTML. It is the time a browser takes to load a page, including all its resources, such as CSS, scripts, images, etc. It's related to user experience more than anything.
How do you determine when a page load is complete?
Comments welcome. Since this is probably the first time this is being done.
If you're collecting the information via Javascript is there any scope to grab some more variables like the CPU type, browser, OS etc?
I can imagine that there may well be bigger contributors to page load time than the ISP used. For instance, what if Orcon users (for whatever reason) have on average older machines?
I think if you're testing "ISP" performance you need to be controlling for everything that is outside the ISP's control?
-- -Michael Fincham System Administrator, Unleash www.unleash.co.nz Phone: 0800 750 250 DDI: 03 978 1223 Mobile: 027 666 4482 _______________________________________________ NZNOG mailing list NZNOG(a)list.waikato.ac.nz http://list.waikato.ac.nz/mailman/listinfo/nznog