[nsp] FW: [nsp] Introducing artificial delay?

From: Sean Knox (sknox@sbcglobal.net)
Date: Thu May 02 2002 - 21:48:06 EDT


Forgot to forward this to the list...

-----Original Message-----
From: Sean Knox [mailto:sknox@sbcglobal.net]
Sent: Wednesday, May 01, 2002 6:56 PM
To: 'Blaz Zupan'
Subject: RE: [nsp] Introducing artificial delay?

NIST produces a linux tool called NISTNet that allows you to introduce
delay, lost packets, and more. I'd personally go with the ipfw solution Blaz
suggested for simplicity's sake.

NISTNet:
http://snad.ncsl.nist.gov/itg/nistnet/

Sean

-----Original Message-----
From: Blaz Zupan [mailto:blaz@amis.net]
Sent: Wednesday, May 01, 2002 8:37 AM
To: ELAW@dr.dk
Cc: cisco-nsp@puck.nether.net
Subject: Re: [nsp] Introducing artificial delay?

> We're sending some of our staff to Korea next month for the World Cup in
> soccer.
> Based on our experiences from the Olympics in Sydney two years ago we'd
like
> to test how their applications perform
> when running from Korea to Denmark.
>
> Any ideas on how we might introduce an artificial delay in our test setup?
> We'll be using Frame Relay or leased line access in Korea, so introducing
> delay between the serial interfaces of two routers could be an option.

You can take a spare PC with two ethernet cards and put FreeBSD on it, then
use dummynet and configure something like this:

        ipfw add pipe 1 ip from A to B
        ipfw pipe 1 config bw 1Mbit/s delay 400ms

This would simulate a 1Mbps link with a 400ms delay. dummynet can also
simulate packet loss. For more information on dummynet you can take a look
at:

        http://info.iet.unipi.it/~luigi/ip_dummynet/



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:11:56 EDT