Re: Accelerator Cache

Well, it's a bit more than that.

In the most common mode, you'd install Squid as a caching proxy. This
means that requests are made in proxy form (GET http://hostname/URI
HTTP/1.0), and Squid resolves the request completely from cache, or by an
IMS to the source, or by a complete request to the source. The browser has
to be explicitely configured to look to the cache, either by the user or by
admin intervention at some point - either way, the browser knows it's
talking to a proxy, and may operate differently than talking direct.

In accelerator mode, the Squid pretends to _be_ the website it's
accelerating. Requests are made in standard form (GET /URI HTTP/1.0), but
Squid still resolves the request in much the same way as before. The
browser thinks that the http_accel Squid _is_ the remote website, it has no
idea it's talking to a proxy of any sort.

[There's also transparent proxy mode, where TCP tricks are done to redirect
requests to the proxy transparently. The browser doesn't know it's talking
to a cache, but the cache is pretty much a normal cache otherwise.]

There are a couple reasonable ways to use http_accel mode. If your web
server is heavyweight (mod_perl, for instance), it may not pay to have it
serving static content. You can either move the static content to a
seperate server, or you could just throw an http_accel Squid in front of
it. The http_accel Squid will cache the static content, leaving the actual
web server to handle primarily dynamic content.

Another use is to provide a local "front" for a remote website. If one
branch of a company has a website that other branches want to see, it may
be more reasonable to throw up an http_accel Squid at the remote sites to
act as a front for the website. That way static content is served from the
local cache, but you don't have to reconfigure browsers to use a proxy.

Later,
scott

----- Original Message -----
From: Dave J Woolley <DJW@bts.co.uk>
To: <squid-users@ircache.net>
Sent: Tuesday, January 11, 2000 3:25 AM
Subject: RE: Accelerator Cache

> > From: Ben Mckellar [SMTP:Ben@hotlinesupport.com]
> >
> > Can someone explain to me Cache and Accelerator for Dummies.... in a
> > practical situation for both,
> >
> An acceleraator is a cache that sits close to the
> server, whereas a normal cache is close to the
> client.
>
Received on Tue Jan 11 2000 - 12:23:31 MST