The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2001-Dec> msg00025



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

MPLS Overhead Questions

  • From: "Gary Blankenship" <garyb@foundrynet.com>
  • Date: Sat, 8 Dec 2001 16:28:14 +0900
  • Importance: Normal

Title: MPLS Overhead Questions

All:

I have a question.  MPLS labels add overhead to the original transmitted data.  Additionally, subsequent labels (VPN) and encapsulations (Martini) add more data.  An original 64 byte packet could have significant percentage (tax).  Example:

A --- (64) --- B --- (68) --- C --- (88) ---- D --à

If device A is the source of a 64 byte Ethernet frame, then device B adds an 802.1Q tag of 2 bytes the LER (Device C) adds L2VPN with Martini encapsulation.  This is an enormous amount of overhead correct (more than 33%)?  We’ve been doing VLAN tagging like this for years.

1.      Does anyone know of the existing issues with VLAN tagging and where I can get some good references? 
2.      Does anyone have any references with the larger issue presented by MPLS (L2 and L3 VPN)?

I already know that to solve the problem you have more bit time slots on the medium; what if all of the wires are the same speed?  Do you rate limit?  If you rate limit how would you do it?  Most vendors, even the ones that are outstanding at MPLS, suck at rate limiting accuracy.  I actually know of real world deployment plans that plan on using 802.3ad link aggregation on LER to LSR links!  Think about that for a moment…

Gary