-----BEGIN PGP SIGNED MESSAGE-----
"J.S.M.Whitaker" writes:
| And a good thing too. Well done whoever had the foresight to
| include that provision.
It's not a clear cut thing, though - e.g. where a better connected
(for JANET users) mirror site exists, people can get to it much more
quickly than (say) via JANET's international links. Better
performance for the punter, less contention for the Fat Pipe, and
potentially money saved on the site's bill too.
For instance, Apple's Star Wars site was our number 2 site (by
download volume) for March, and Internet Explorer 5 downloads were at
number 7. TUCOWS comes in at 16, and cdrom.com at 19. Details at:
<URL:http://wwwcache.ja.net/Statistics/1999/03/report/sla.1999-03.html>
See Dave Beckett's slides from Networkshop for some possibilities for
how the new JANET Mirror Service (HENSA++ :-) might interact with
the (still legal for the moment!) caching infrastructure ...
<URL:http://www.cs.ukc.ac.uk/people/staff/djb1/talks/networkshop27/>
The bottom line is that where we have a big mirror site, this could
participate in the caching infrastructure by declaring itself as
holding the various URLs which may be used to reach the resources it
knows about - using protocols like ICP and Cache Digests. This isn't
the same thing as cache operators deciding to redirect traffic for
particular domains - viz. blocking or banning.
The Internet as a whole needs to do a lot more work on ways to get the
punter fetching stuff from topologically nearby copies of widely
replicated resources - this is just the beginning!
Martin
-----BEGIN PGP SIGNATURE-----
Version: 2.6.3i
Charset: noconv
iQCVAwUBNxyGbtZdpXZXTSjhAQGQmQQApwOBrWTNfRfzHsGOSWymsyVSAeM5kaQe
NDS2HcrjY8HRWC4xO/CSJxEhk9DbKPTynO1M83X+Mab/10R2Pu+VHjU9NBl4atsC
9KHZceFEs5+0L40QlPo+4oLy21VnzQFx5iF7l2X8tK+7zsUTmnWg98Zl0qjKqwb2
oq5zRncPWGM=
=kb6F
-----END PGP SIGNATURE-----
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|