[ previous ] [ next ] [ threads ]
 
 From:  Chris Buechler <cbuechler at gmail dot com>
 Cc:  m0n0wall at lists dot m0n0 dot ch
 Subject:  Re: [m0n0wall] Re: Hardware Failover?
 Date:  Mon, 30 Jan 2006 01:05:12 -0500
On 1/28/06, Rui Correia <rds underscore correia at mailshack dot com> wrote:
>
> If you're running m0n0 on a Soekris/WRAP I would warn you that pfSense is muuuch
> slower than m0n0.
> While it's true that pfSense has CARP and that it actually works, running it on
> a net4501 will make you go crazy because it is really slow.
> I know it because I've been using pfSense at home on a net4521 until a couple of
> weeks ago.
>

To be fair, that really isn't the case anymore.  (and I think m0n0wall
is the only option of the two right now for a production environment
where uptime is critical, I'm not taking sides, I'm just "keeping it
real")

Average page load times on a 4501 were 15-20 seconds, sometimes more,
up until a major change either in the first beta, or in one of the
subsequent release candidates for beta 2.  Now, you can load pages
side by side to compare (or time it with Fasterfox, that's what I did)
and it's slightly slower than m0n0wall now.  About 3 second average
page load time on m0n0wall on a 4501, and 3.5 seconds on pfsense. 
some pages take much longer on both, but there is not a huge
difference between the two in page load times.

That's just the GUI though (which is likely what he was referring to
above, but regardless...).  With FreeBSD 6.0 (pfsense) vs. 4.x
(m0n0wall), on uniprocessor systems, there is no comparison on network
throughput on low end hardware.  6.0 is roughly 40% slower, depending
on the circumstances.  With SMP systems, 6.0 is faster.

But that's just a consequence of having to abandon FreeBSD 4.x.  The
next version of m0n0wall will see the same drop off in performance in
network throughput.

-Chris