Re: Multisuperfetcha: load balancing the HTML

On Fri, Mar 10, 2006 at 05:05:10AM -0500, <dolphinling>:
~> What happens if one of the servers sends different content than the others?

That's a problem of the website developer. He has to be sure that the mirror
is really a mirror.
A first check would be to verify that the size of the three different
downloads is the same.

On Fri, Mar 10, 2006 at 10:39:14AM +0100, <Laurens Holst>:
~> How about first of all making it valid XML?
~>
~> <a href="http://www.python.org/ http://www.python.mirror1.org
~> http://www.python.mirror2.org/">Python Powered</a>
~>
~> Would be a good start. But then, that would conflict with people having
~> spaces in their URLs, which is done quite frequently.

What about:
<a href="http://url1||http://url2||http://url3">...</a>
or something like that.

On Fri, Mar 10, 2006 at 11:26:05AM -0000, <Dave Woolley>:
~> URIs are standardised by IETF, not W3C (you basically have a three
~> part URL, not three separate URLs).

ok.

~> Also, the above violates the SGML and XML core syntax so is never
~> going to be possible.

Ok, let's find a good solution then.

~> Generally, servers have more bandwidth than clients, and this is
~> only of benefit for uncached resources where the servers are the choke
~> points.

This is not always true, for example if the client uses multi inet-gw it 
will have a great advantage. The multigw is described here:
http://idiki.dyne.org/wiki/Multisuperfetcha

~> There is a problem in synchronising the parts.

This is true if we are talking about a html page, but what about:
<a href="http://url1/distro.iso", "http://url2/distro.iso", "http://url3/distro.iso"">...</a>

If they are mirrors, they are already synchronized.

Best regards.
-- 
:wq!
"I don't know nothing" The One Who reached the Thinking Matter   '.'

[ Alpt --- Freaknet Medialab ]
[ GPG Key ID 441CF0EE ]
[ Key fingerprint = 8B02 26E8 831A 7BB9 81A9  5277 BFF8 037E 441C F0EE ]

Received on Saturday, 11 March 2006 05:23:00 UTC