The MPLS-OPS Archive

Cell Relay Retreat>MPLS-OPS Archive>month:2003-May> msg00071



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

RE: Traffic Engineering

  • From: Jean Philippe Vasseur <jvasseur@cisco.com>
  • Date: Mon, 12 May 2003 13:09:14 -0400
  • Cc: "Rudisill, Amelia" <rudisila@ncr.disa.mil>, <mpls-ops@mplsrc.com>
  • Resent-Date: Mon, 12 May 2003 13:54:00 -0400
  • To: Sebastien.Spas@alcatel.be
  • X-Sender: jvasseur@paris.cisco.com

Hi,

At 09:56 AM 5/12/2003 +0200, Sebastien.Spas@alcatel.be wrote:
>Hi Amelia,
>
>What I saw the most was RSVP-TE directly on the P routers, most of the time
>with explicit LSPs.
>It looks like fast reroute is not highly implemented yet, but it's becoming
>mandatory with DiffServ MPLS for voice/video LSPs.
>Concerning the protection methods,

Quite untrue. There are mix of networks running distributed CSPF and 
off-line computation. Actually probably more networks run distributed CSPF 
for TE LSP path computation. Anyway both approach are completely valid.
As far as FRR is concerned, FRR bypass is now largely deployed on several 
small, medium and large networks.

JP.

>1:N (fast reroute by-pass LSP) is the most used, maybe because it's the only
>one supported by cisco.
>Administative cost is higher than 1:1 method (you must manually create and
>maintain all the by-pass LSPs), but bandwidth cost is lower since it's
>sharing protection. Scalability is better of course.
>Assumption is made for this approach than only 1 network element (node or
>link) can be down at one time. It's quite realistic. This mechanism can also
>be linked with SRLG approach.
>
>1:1 (fast reroute detour LSP) is not used much. Administrative cost is very
>low (simply need to set a fast-reroute tag on your LSP object). Bandwidth
>cost is very higher (because no sharing of backup capaciy), and scalability
>is lower (for an LSP going through 5 nodes, 4 additional LSPs will be
>created, signalled and reserved). You can still do bandwidth sharing by
>specifying the detour LSP capacity to be zero.
>
>kr,
>sebastien.
>
>-----Original Message-----
>From: Rudisill, Amelia [mailto:rudisila@ncr.disa.mil]
>Sent: vendredi 9 mai 2003 19:49
>To: 'mpls-ops@mplsrc.com'
>Subject: [MPLS-OPS]: Traffic Engineering
>
>
>Hello,
>
>Anyone running MPLS-TE on their network and what capacity?  If so, are you
>running RSVP-TE over LDP or just RSVP-TE on the P routers?  Are most
>providers implementing Fast Reroute? Cisco only implements 1:N method.  Are
>most Service Providers implementing 1:N or 1:1? What is the primary
>protection method implemented, Link or node protection?  Can someone provide
>some advantageous and disadvantageous of each?
>
>Thanks,
>
>Amelia
>
>-------
>The MPLS-OPS Mailing List
>Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
>Archive: http://www.mplsrc.com/mpls-ops_archive.shtml
>
>-------
>The MPLS-OPS Mailing List
>Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
>Archive: http://www.mplsrc.com/mpls-ops_archive.shtml

-------
The MPLS-OPS Mailing List
Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
Archive: http://www.mplsrc.com/mpls-ops_archive.shtml