[techtalk] HELP! Webserver compromised?!?

[Mary Gardiner]

On Thu, May 03, 2001 at 08:46:10AM -0400, Michelle Murrain wrote:
> I know about mod_proxy, and the general idea of proxy servers, etc. But how 
> does it make any sense in terms of saving bandwidth, for me, for example, to 
> relay through a different server, to fetch web pages? The web pages and 
> associated files are the same size, whether I get them through another server 
> or on my own, and so I'm using the same bandwidth to retrieve the files. 
> Further, it would inevitably be slower to get those pages, since I'm going 
> through another server. 
> 
> What am I missing?

It's worth pointing out in more detail that a lot of the major development work on
proxies has been done by people who aren't from the US. The reason is that the
most often viewed webpages are from the US and the cross continental cables are
a bandwidth bottleneck.

While I'm on cable at home, and my theorectical bandwidth is 64K/s I don't get that
from the US debian site, I get 2-10. I get that from the Australia mirrors though.

Hence if I can use a caching proxy for often-viewed and comparitively seldom-changed
webpages, and particularly for their images, which tend to change the least, I can
save a lot of time.

Martin Pool, at Linuxcare Australia is even developing a proxying method based on
rsync, so you can request only the changed data in a page based on exchanage of
checksums via the rsync algorithm. This is based on the idea that many pages, like
Slashdot have a great deal of tables and so on for formatting purposes (they might
consititute above a third of the page for example) that very rarely change, whereas
the data may change more rapidly.

Mary.

-- 
Mary Gardiner
<mary at puzzling.org>
GPG Key ID: 77625870