Re: Square 1

On Sun, 28 Dec 1997, Paul Matthews wrote:

> 1. The term "proxy" is a bit ambiguous. Does it mean to conceal
> the real host (as in firewalling)? Or is this usage meant to show
> that it relieves one or more hosts from excess workload? [I really
> did not understand, but I assumed the latter.]

"Proxy" is simply, "on behalf" - which means it applies in any case
wherein one entity performs an action on behalf of another entity; in
this instance, the Squid software retrieves Web objects on behalf of some
Web browser (client) rather than the browser going to retrieve the object
itself.

The answer to *why* this may be used is entirely up to the network
administrators - for example, it may be used to purposely conceal the real
clients (as in firewalling) and/or it may be used to provide the benefits
of a shared cache to a large(r) population of users (such as reducing
costs, lowering traffic on WAN links, &c.).

Yes, it would ultimately relieve the target host from the load of
repetitive queries but this isn't a primary intention (the benefits are
aimed at the requesting end of the transaction, not the answering end).

> 2. When you say "server" do you include a network gateway machine?

It is a "server" in the "client/server" model - that is, you run a browser
(eg., Netscape) on your machine .. this browser is the client software;
somewhere on the network (be it your own machine, a shared server, a
dedicated server or some firewall box) runs the Squid software .. this is
the server software to which your browser connects to request Web objects.

Of course, in acting as a "proxy" the Squid software must then submit your
query to someone else (either either proxy servers or directly to the
target site) and in that case the Squid software becomes the client with
the destination of the request being the server.

> Or does squid have to run on a web server?

No - only the "cachemgr.cgi" component is said to require a Web server in
order to provide an easier view of current Squid operations; the Squid
server itself doesn't require a Web server at all.

> I run a Linux box essentially as an ISDN router. Will squid function on
> the gateway/router to cache http and ftp traffic between my LAN of 5
> workstations and the Internet?

Should do, yes.

[...skipped a few questions...]
> 5. Can you offer any general warnings/suggestions/things to look
> out for?

Run it up on a "test" box first - play with it, get used to how Squid
works and how it might best be configured for your environment. Then
start to carefully install it onto your ISDN router box if you so wish.

Cheers..

dave
Received on Sun Dec 28 1997 - 07:52:17 MST