Re: Performance issues on WRAP

If you want to stay in that realm (low power, single board system), I'd
recommend looking at the Soekris 6501. It has an Atom CPU and quad gig
ethernet. Should be able to handle several hundred megs of routing, if not
fully saturate a gig.

--  James L. Lauser
    [email protected]
    http://jlauser.net/


On Mon, Apr 23, 2012 at 09:30, Stefan Ott <[email protected]> wrote:

> On Sun, Apr 22, 2012 at 06:55, Chris Buechler <[email protected]> wrote:
> > On Sat, Apr 21, 2012 at 11:08 PM, Stefan Ott <[email protected]> wrote:
> >> Hey
> >>
> >> I recently noticed a strange behaviour with m0n0wall 1.33 running on a
> >> PC Engines WRAP. It seems that whenever the network throughput (tried
> >> with http and ftp) gets close to my connection's maximum (which is
> >> around 25 Mbps) the system becomes extremely unresponsive and won't do
> >> anything else anymore - DNS lookups fail and even wifi connections get
> >> terminated. A quick look at the CPU load graph shows that it's very
> >> close to 100%.
> >>
> >
> > At 25 Mbps, you're pushing everything you're going to get through a
> > WRAP. Things are going to get unresponsive at that point, you're going
> > to start dropping packets, etc. Enabling polling won't really help,
> > it'll at least keep userland processes on the system from being
> > overwhelmed by interrupt load, but it will slightly reduce the amount
> > of throughput you can get and increase latency. Really need a faster
> > box than a 233 or 266 MHz Geode at 25+ Mbps.
>
> :(
>
> But thanks a lot for the answers, I honestly didn't expect some minor
> routing and NATting to use that much CPU power. Guess it's time to go
> shopping then!
>
> On a related note, does anyone happen to know whether the ALIX boards
> (500 MHz Geode) have enough power to be reasonably future proof? Or is
> it a simple linear thing, i.e. I'll have the same issues on an ALIX
> once I hit 50 Mbps?
>
> cheers
> --
> Stefan Ott
> http://www.ott.net/
>
> "You are not Grey Squirrel?"
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
>
>