From: Keane, James (James.Keane@agriculture.gov.ie)
Date: Thu Jun 09 2005 - 11:43:04 GMT-3
well you could always set up a 2mb connection using a serial DB60 back to back.
Then run it at MAX with WRED set up in the default class referencing a child policy which does the shaping.
now you need to manipulate the random-detect to always random drop regardless of queue size, if you set the minimum size to 0 I believe now you are in business, you may need to put in a low bandwidth statement to fool MQC.
Whats the general opinion ?
Interesting to lab it.
One thing, trying to recreate the .25 second delay for hitting the satellite and the .25 for the return will be a lot tougher ... unless you have a very long serial cable (he he)
-----Original Message-----
From: Zamberlan, Diego [mailto:dzamberlan@verisign.com]
Sent: 09 June 2005 15:15
To: Daniel Sheedy; ccielab@groupstudy.com
Subject: RE: satellite test lab
IP over satellite links? That sounds soooo 70's... :)
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Daniel Sheedy
Sent: Thursday, June 09, 2005 5:01 AM
To: ccielab@groupstudy.com
Subject: OT: satellite test lab
Hi guys,
I have an OT that seems a bit interesting.
A guy at work is writing a paper for his Master's thesis about
Satellite link communications. He has to build a demonstration of the
link, and we are looking for something where we can simulate a 2meg link
with random packet loss and traffic shaping. The traffic shaping is no
problem... but how does one introduce random packet loss into a link?
Has anyone needed to do this before? Also, does anyone have a
recommendation for a detailed book covering TCP/IP? Thanks
Daniel Sheedy
This archive was generated by hypermail 2.1.4 : Wed Jul 06 2005 - 14:43:41 GMT-3