Re: Accessing wwwcache.hensa using squid (fwd)

From: Oskar Pearson <>
Date: Thu, 24 Oct 1996 08:45:02 +0200 (GMT)

Hi all

> >about 2.8 million requests per day. If you know of a Squid server
> >successfully handling this kind of load we would be interested to hear
> >about it.
> Did the architecture of the Netscape proxy-server change in version 2? Has
> it become a single process server which caches metadata in memory just
> like Harvest/Squid?
> I've seen version 1 of the Netscape proxy-server and its architecture
> looks like a CERN daemon (no caching of metadata in memory, URL's directly
> translated into pathnames without hashing, etc.) but with preforked
> children. The problem I had with version 1 was that performance collapsed
> for large access lists (>1000 IP numbers).

Do you think that this is why they had the "one second delay" when they
tried squid? I know that they restrict access to the cache to "" domain,
so either they do it by reverse looking up every IP, or they have a large
IP based ACL list.

The other possible reason for it is that it could be because we use the
"dnsserver" proccesses? I have had a quick look at the source, and it seems
that it is essentially a wrapper that caches (I am not sure how - it seems
like each "sub-program" (ie squid/ftpget) is responsible for accessing
the dns cache directly?) I don't see how it would be inefficient, unless
they are using a "reverse-dns" based ACL... they would then have to do
2 lookups per access, which could take a while...

Received on Wed Oct 23 1996 - 23:45:21 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:33:20 MST