Most traffic shaping is between the WAN and the LAN Interface. Traffic 
shaping can also only be done with traffic leaving the interface.(WAN 
Interface=Upload Shaping, LAN Interface=Download Shaping)

Most of the time, you set priorities for traffic, so for example. VoIP 
always has bandwidth available, while an FTP upload can be dropped at 
anytime.

I don't know how m0n0wall does traffic shaping, but for pfSense this 
document explained it pretty well.

http://www.openbsd.org/faq/pf/queueing.html

Adam

Michael Pope wrote:
> I'm trying to understand how to implement traffic shaping. We haven't had
> much need for it in the past.
>
> As I understand it, traffic shaping doesn't give much benefit if it isn't
> implemented throughout the network. So, traffic shaping could provide
> benefits within a network that has multiple subnets. (For example, a
> multi-site network where the sites are connected using either site-to-site
> VPN or a dedicated link).
>
> But, I'm not sure about the benefits of traffic shaping at the network
> perimeter. How could it help your traffic if that traffic is immediately
> off-loaded into a network that might/might not have shaping implemented? It
> almost seems like the high-priority traffic is getting into the HOV lane
> right before the off-ramp.
>
> Can someone point me to an explanation or provide it themselves?
>
> Thanks,
>
> Michael
>
>