Re: Caching cgi hits

From: Henrik Nordstrom <hno@dont-contact.us>
Date: Mon, 24 Feb 1997 23:23:52 +0100

Hi Thomas,

One thing to remember when you think about a accelerator
is that it is in no way different from a cache (only the
content that it caches is different). The same rules apply.

The main reason why you may want to change the URLs to
point to your real server port (81) is performace. There
is a performace loss in going thought the accelerator
for non-cacheable pages, and it causes a higher load on
your machine since on more process is involved eating
your hardware resources.

In teory performance is the only reason for changing the
URLs to the real server port, but in practice it helps
to aviod misconfigurations in Squid, or caching of pages
that should not be cached but they claim that they might
be (by setting Last-Modifier or Expires headers).

There are some other cases where you don't want to go
thought the accelerator, and that is if the source of
the request is important (for access control or whatever).
(this should be depreated.. violates all cache thinking)

In any case, you should make sure all your uncacheable
pages set the Expires: header correctly. Caches are
getting widespread use these days, and most of them
uses a default TTL if Expires: is not set.

---
Henrik Nordström
ps. Questions should be sent to Squid-users since they
might be of general interest, and others might answer.
TBRIGLIA.US.ORACLE.COM wrote:
> 
> Hi Henrick,
> 
> Thank you very much for the reply! Sounds like you have some
> in depth experience on the caching technology.
> 
> I will be implementing the Squid cache this week as an
> accelerator, and may occaisonally need some questions
> answered. I may contact you sometime if I get stuck
> and need a question answered.
>
> Right now I do have one question. Right now my main httpd
> server runs on port 80. I plan on moving it to port 81, and
> set the accelerator to port 80. Do I need to do any
> modifications to any of the existing URLs in my current
> document home? The Squid documentation under the "httpd
> accelerator" section states;
> 
> "In such a configuration, the web administrator renames all 
> non-cacheable URLs to the httpd's port (81)".
> 
> I plan on removing the rule in the conf file that makes
> /cgi-bin/ hits ignored so that the cache will try to store
> cgi-bin hits. With this in mind, do I understand that the
> above statement means that if there were certain URLs that
> I do NOT want to cache, I should change their URL so that
> it includes ":81" the port of where the true httpd server
> now listens?
> 
> Thanks in advance for any light you can shed on this.
> 
> 
> Regards,
> 
> 
> 
> -----------------------------------------------------------------------------
> 
> Thomas D. Briglia  voice:415-506-8212   pager:1-800-6-PAGEME
> Internal Systems Engineering Services
> Oracle World Wide Customer Support                      Viva BSD!!!
Received on Mon Feb 24 1997 - 14:51:57 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:34:32 MST