On 8/6/07, Duncan Drury <DDrury at christian dash aid dot org> wrote:
> I am using m0n0wall to model the network conditions that my
> organisation's overseas offices experience.  This allows us to test how
> well different applications perform under those conditions and make
> adjustments accordingly.  I thought I'd share how I am doing this in
> case it is useful to others, and also to see if anyone sees any problems
> with my method.
>
> For each office I am using www.speedtest.net to get an approximate
> measurement of bandwidth both up and down, and the delay.  I then take
> the figures I have and create two pipes in m0n0 for each office - one
> with the download bandwidth and one with the upload bandwidth.
>
> I then create a rule for each of these pipes - the download rule is for
> incoming traffic on the WAN for any protocol, any port and to and from
> any host.  This has the download pipe applied to it.
> I create a similar rule for uploading except here the rule is for
> outgoing traffic on the WAN.
>
> I then activate only the two rules for the particular office I want to
> model.
>
> So far this seems to be effective.  Has anyone got any suggestions for
> how I might improve this?  I'm particularly interested in any ideas on
> how to better gather the information on network conditions in our
> offices.  Does anyone see any flaws in my use of pipes and rules?
>

You could further edit your pipe to introduce delay and packet loss if desired.

db