The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2003-Nov> msg00167



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

on documenting ECMP (was on the mpls oam framework)

  • From: "Naidu, Venkata" <Venkata.Naidu@Marconi.com>
  • Date: Tue, 18 Nov 2003 18:57:49 -0500
  • Cc: "'David Allan'" <dallan@nortelnetworks.com>, "'tnadeau@cisco.com'" <tnadeau@cisco.com>, mpls@UU.NET

Peter,

-> Forgive me for being simple-minded:

  Forgive me for such a general explanation.

-> My question is: if I use RSVP-TE to be guaranteed that 10 
-> Mb/s of bandwidth is reserved for my aggregate traffic, 
-> would I ever want to use ECMP splits in the end-to-end path?

  Not required. But if ECMP is enabled you never know.

-> I understand that that is theoretically possible within 
-> agreed-upon limitations. But are these limitations narrow 
-> enough to make this a realistic scenario. Or do I just have 
-> to reserve 10 Mb/s along each of the multiple paths?

  No. No need to reserve 10 Mb/s along each of the multiple path.

  Some ECMP models are control plane directed with a prior
  knowledge of far away hops and current load on bottle-neck 
  links. If, for example, there are two different paths to
  reach the destination. Each of those paths have different
  bottle-neck links and the demand for some links are high
  along a path then ECMP can split traffic proportionately.

  In such a case reserving 10 Mb/s is not advised. Reserving
  10Mb/s on every multiple path is an overkill. Some vendors
  can do so for resiliency. I never heard of such.

Venkata.