I was doing some testing and I was able to server around 550 requests/sec
with squid, and about 1100 with apache, both under linux.  Squid with
freeBSD was maxing out at around 330 requests/sec, but I was getting an
error that said it couldn't allocate more memory for mbufs or something like
that.  

Why is the default FD_SETSIZE in the kernel set so low?  Most systems now
can handle much more that systems built 2 years ago.  When trying to build a
high performance server, it makes it extremely annoying.  I had to do modify
the kernel to get more concurrency out of qmail also.  FreeBSD allows you to
change this on the fly with sysctl, but it seems there are some other
bottlenecks that you can only fix with a recompile.

ARGH!  

> -----Original Message-----
> From: Carl Wilhelm Soderstrom [mailto:chrome at real-time.com]
> Sent: Thursday, March 29, 2001 2:49 PM
> To: tclug-list at mn-linux.org
> Subject: Re: [TCLUG] reverse proxy/web accelerators
> 
> 
> > Besides Squid, are there any other high performance reverse 
> proxies/cache
> > servers for Linux?  How do they compare with squid?
> 
> I heard once that Apache+some proxying module, was a better cache than
> squid. It was a while ago, and I don't remember much more than that. 
> 
> Squid has probably gotten better, tho. :)
> 
> Carl Soderstrom
> -- 
> Network Engineer
> Real-Time Enterprises
> (952) 943-8700
> _______________________________________________
> tclug-list mailing list
> tclug-list at mn-linux.org
> https://mailman.mn-linux.org/mailman/listinfo/tclug-list
>