RE: Asymmetric Routing

From: Mike Williams (ccie2be@swbell.net)
Date: Mon Nov 17 2003 - 00:04:41 GMT-3


Danny,

If a requirement states that R1 must send all packets for R3 via R2,
then ensure that all packets from R1 to R3 go via R2. Don't assume
anything else. Don't assume that you have to make all return traffic
take any specific path. Unless a requirement expliciticly says that
it's needed, don't waste time and take a chances on messing things up by
trying to fulfill something that's not even a requirement.

Mike W.

Danny.Andaluz@triaton-na.com
Sent by: nobody@groupstudy.com
11/15/2003 12:08 AM
Please respond to Danny.Andaluz

 
        To: ccielab@groupstudy.com
        cc: (bcc: Donny MATEO/ADPC/ASIA/BANQUE_INDOSUEZ/FR)
        Subject: Asymmetric Routing

Hey, Group.

I had a question about asymmetric routing. Here's a simple diagram:

http://home.comcast.net/~practice.lab.examples/asymmetricrouting.jpg
<http://home.comcast.net/~practice.lab.examples/asymmetricrouting.jpg>

If a requirement states that R1 must send all packets to R3 via R2, is
it safe to assume that the traffic has to come back from R3 the same
way? Is this the case even if it is not explicitly stated in the
requirement that
it
must? Would it be ok if the return traffic came back via R4? I know
this is probably not best practice, but for a lab, would it matter? I'm

thinking
traffic should be sent and returned the most optimal way, but in this
case,
the paths from R3 to R1 have the same metric.

Thanks,
Danny



This archive was generated by hypermail 2.1.4 : Fri Dec 12 2003 - 12:29:13 GMT-3