From: Scott Morris (swm@emanon.com)
Date: Mon Aug 21 2006 - 19:44:43 ART
So if you look at your routing table on any given router, how many routes do
you have? Now, how many different next-hops do you have?
If you could lump your entire routing table into a smaller table with ISP-1,
ISP-2, ISP-3, and Internal as your only four viable choices for any
information, wouldn't that be more efficient?
It's all about speed and efficiency. Although I do appreciate you not
wanting to believe everything you read as propaganda! :)
Scott
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of Tim
Sent: Monday, August 21, 2006 10:52 AM
To: ccielab@groupstudy.com
Subject: MPLS speed -- switching vs IP routing
Hi guys,
I'm reading lots of different material on MPLS and all these sources talk
about how one of the benefits of MPLS is speed.
They all say that label swapping is faster than route table lookup. They
also talk about other benefits to using MPLS but I'm wondering how much of a
benefit is speed.
I'm not sure I'm buying this. My hunch is that this speed "benefit" is
being propagated mostly by MARKETING people, not technical engineers.
Why would doing a label lookup instead of a route table lookup be any
faster?
And, even if we accept the premise that switching labeled packets through a
network is faster than plain old ip routing, how much faster is this really?
Are there some large networks where the difference in speed makes a big
difference in overall performance?
Have there been any studies that quantify how much faster using MPLS is
versus legacy IP routing?
So, basically, I'm trying to get some perspective on this speed thing to
understand if this is negligible benefit, a minor benefit or a major benefit
and under what scenarios would result in which level of benefit.
Thanks, Tim
This archive was generated by hypermail 2.1.4 : Fri Sep 01 2006 - 15:41:58 ART