So assuming all other things are equal (a wildly optimistic assumption but stay with me), how much of a difference to web page load time would it be between.

A) connecting to a web page hosted on a server across the street
B) connecting to a web page hosted on a server on the other side of
the world.

I never did physics at school but my understanding is that electrons travel on a wire at the speed of light or thereabouts (300,000 km / sec) and the circumference of the earth at the widest point is 40,000 km so with perfect hardware (you'll need to imagine that) a request could go to the farthest point of the earth and come back in 1/7th of a second.

So "distance" per se is not relevant, theoretically.

Yet in reality anything travelling across the world wide web has to hop and bounce from one network to another. My understanding is that these "network hops" (I'm sure there is a better word but I dont know it) are what would create the real difference in connection time between scenario A and scenario B above.

Having said that.....

How much of a difference is it really?
Is this geographical sand in the gears really something that can be perceived by the end user?

This article does a terrific job of testing out the effect of pinging a set of 10 different websites from 3 different locations (Germany/India/USA) and testing the outcome.
http://www.webdevhub.net/website-opt...te-performance

From this it's clear that there is some time difference BUT the difference
in response time from the very fastest to the very longest is only 500 ms (half a second).

My conclusion from this is that when it comes to making a choice of server hosting company the "geographic distance" is a factor so minor as to be irrelevant in the equation.

Of course there are other factors like potential SEO impacts, cost, performance, physical risk etc.
I know about those things and I DON'T WISH TO DISCUSS THOSE IN THIS THREAD.
My question relates SOLELY to the impact of distance on load time performance.

Any comments?

http://tycoontalk.freelancer.com/web...g-company.html

And BTW "hop" is actually the terminolgy used to bescribe the packet's journey between two nodes.

...with the speed and sizes of the fibre "pipes" that run around the world now latency is not an issue.
Fifteen or twenty years ago when most of the intercontinental communication network was run in copper cables, latency was an issue mainly due to the physical connections and the equipment used. (the speed of electrons in copper and the speed of light in fibre are approximately equal because there is a ~0.66r reduction in the speed of light through fibre as opposed to the speed of light in a vacuum.

Now, because the technology has advanced, equipment is much faster and no longer relies on physical contact to route the signals, latency is much less of a concern and is measured in nano seconds rather than the milli seconds of the late 1990s equipment.

Thanks. That's what I needed.
Not that I'm doubting at all, but is there a canonical reference source
that I can refer doubting clients to?
We always use major hosting companies in the USA even though a lot of the user base is in Asia.

I just needed something to disprove the network latency the argument.

I'll see what I can dig up for you.

Just to follow up on this,

if network latency is close to zero, how do guys
like these make a buck by selling CDN hosting
at premium rates?

http://vps.net/product/cdn/

The YSlow analytic Addon tool says this about CDN:

User proximity to web servers impacts response times. Deploying content across multiple geographically dispersed servers helps users perceive that pages are loading faster.

Why do people buy diet pills?

Why do people buy hair restorer?

Why do people buy "generic cialis"?

Simple! They think it will make a difference



Do some timing checks from around the world to your server with traceroute , http://traceroute.org/

Yes, geographic distance makes a difference. You need to keep in mind that every pageload is pulling in multiple resources, and it's important to realize that a human will notice the difference between a 20ms latency and a 300ms latency.

It can be the difference between an entire page snapping into view as soon as the link is clicked, and the page taking half a second to load up, including all images, css and javascript.

I live in Europe, but am from the States. Many of my customers are in the US, and we usually host customer projects in the US. When I'm in the States with a 20ms latency to our hosting company, I see a noticeable difference in pageload speeds, compared to when I view the same sites while in Europe.

On the bright side, I can recommend an excellent hosting company that has servers in the US (Texas), Europe (Amsterdam) and has just this week opened up shop in Asia (Singapore). Their prices and service are excellent -we've been hosting with them for 2 years and have absolutely no complaints.

They're called WebFaction and are definitely worth a look if you'd like to be able to work with one (UK-based) company with servers around the world. It's completely transparent where your server is - you could set one up in the US, one in Europe, and one in Asia, and host projects closest to the customer, all working with one company.

I'd encourage you to give them a look.

Thanks for reply texinwien.

Interesting that not everyone agrees with you re latency.

I may take a look at that hosting company some time in the future.