RE: Caching Algorithms

Well, usually, but not always. For example, there are sites that set their
TTL to 0 in order to boost their hit rate. Others, like Microsoft (at the
last time I heard) had set the TTL to 0 for all their pages, including
buttons and graphics. I guess they get kickbacks based on number of bytes
sent :-)

> -----Original Message-----
> From: Rodney Holm [SMTP:rodneyh@apexxtech.com]
> Sent: Thursday, July 30, 1998 2:43 PM
> To: squid-users@ircache.net
> Subject: Re: Caching Algorithms
>
> Showing a page that the cache knows is incorrect does not seem to be a
> very
> good idea. This could potentially lead to someone using the information
> on
> the cached page as up-to-date. For instance on a stock quote site, the
> user
> could choose to sell their stock based upon this incorrect quote. This is
> but
> one example of how serving a stale cached copy could cause a serious
> problem.
>
> If a site tells a cache that it doesn't want to be cached or tells a cache
> that
> it wants to expire after a certain period of time, there is probably a
> good
> reason for it.
>
> Jordan Mendelson wrote:
> >
> > Curious, it seems that Squid, when receiving a web page with the
> no-cache or
> > expires tags will not send it back to the client after it expires, but I
> was
> > curious....
> >
> > If the site goes down and we have that cached copy in memory or on disk,
> why
> > not send it back instead of the "read timeout" message. This happens (a
> lot)
> > with http://slashdot.org. It seems to get overloaded a lot, the link
> goes
> > down, etc.
> >
> > This is how Netscape's internal cache works (if it can't contact the
> host,
> > it displays the previously cached copy even if the page instructed not
> to
> > cache or to expire).
> >
> > Jordan
> >
> > --
> > Jordan Mendelson : http://jordy.wserv.com
> > Web Services, Inc. : http://www.wserv.com
>
> --
> --------------------------------------------------------------------------
> -----
> Rodney D. Holm rodneyh@apexxtech.com
> Apexx Technology, Inc. http://www.apexxtech.com
> --------------------------------------------------------------------------
> -----
Received on Thu Jul 30 1998 - 12:52:28 MDT