General Questions

Hi all,

We are looking at m0n0wall to use in a system which has a number
of WRAP boards running an older version of Pebble Linux.  m0n0wall
looks like a great replacement, and the level of participation and
support are impressive.

Because these WRAP boxes are used in an entirely wireless system
with 240+ end users and 40+ AP's linked together with multi-hop
backbone 802.11/a links, iperf is the tool of choice to test the links
from the WRAPs to various points in the system.  Multi-hop wireless
can be difficult to diagnose without an actual traffic testing tool.

I couldn't find any mention of iperf in the Fora or in the documentation,
and I apologize if I overlooked it.  If not iperf, is there another traffic
testing tool available within the m0n0wall package?  If neither, was I
supposed to post this in the Feature Requests?

On behalf of the tech group of the La Canada Wireless Association:

http://www.lcwireless.net

TIA,


Nick Espeset

No there isn't. But you really don't want to test using one end of the test on the device itself because it'll skew your test results.

Testing that way doesn't provide an accurate picture of how real traffic would perform because it's nothing like what will actually happen - your firewall won't be the endpoint for that traffic, and it won't be spending resources generating and responding to that traffic.

Well, actually, in our topology these boxes sit at the end of backbone links, between backbone and access points.  What we are testing isn't network throughput, but the radio's bandwidth.

In a conventional wired network, bandwidth doesn't change, even if throughput changes.  Unless there is a system failure, i.e. a NIC is failing in a noisy manner, or a bad cable, etc. one doesn't generally question whether there really is a 100 Mbps ethernet link or not.

In a wireless system with long paths and multiple backbone hops, the actual 'pipe diameter' changes dynamically based on, predominantly, environmental factors.  Unfortunately, actual throughput is not one of the things that can be determined by polling the radios.  They will report signal-to-noise, overall signal levels and such, but they can't tell us, for example, that due to a temperature inversion, a multipath signal is creating phase interference, choking the bandwidth.  Limiting factors include wind induced mechanical jitter, variating Fresnel interference, and - unfortunately - many others. 

Our experience is that iperf is extremely valuable, since it can be used to quickly and remotely check the point to point radio performance in the overall system.  I hope that helps explain our interest in iperf.

In that case it makes more sense. You should be able to compile iperf on a FreeBSD 4.11 box and copy the binary over to a running system, or make an image that includes it. Start a new thread under development if you have questions related to that.