RE: MPLS load balancing

hi Ray

could u plz explain how u resolved this problem

Regards

Asim

>From: Raymond Cheung

>Reply-To: rcheung@slb.com

>To: raszuk@cisco.com, Ayush Sharma

>CC: tpittman@frontrunner.eu.com, "''Peter J. Welcher' '" , mpls-ops@mplsrc.com

>Subject: RE: MPLS load balancing

>Date: Fri, 28 Dec 2001 00:25:20 -0600

>I've made them working with different IOS 12.2(2)T. I used 12.2(4)T1 for

>7206VXR and 12.0(19)ST1 for GSR.

>Since this is a new feature, I'm just trying to see if anyone could help. I

>doubt TAC would help you on lab testing, they probably would, but I'm glad I

>just make it working by myself.

>regards,

>Raymond

>-----Original Message-----

>From: Robert Raszuk [mailto:raszuk@cisco.com]

>Sent: Friday, December 21, 2001 10:01 AM

>To: Ayush Sharma

>Cc: 'Raymond Cheung '; 'tpittman@frontrunner.eu.com '; ''Peter J.

>Welcher' '; 'mpls-ops@mplsrc.com '

>Subject: Re: MPLS load balancing

>In the mean time when you assing different RD - RR simply does not

>compare the routes and send vpnv4 updates received from PE1 as well as

>from PE2 for the same prefix to PE3 just fine. Something must be

>misconfigured as this works just fine.

>Complaining that this is broken will not help to troubleshoot the issue.

>Pls open a TAC case and they will help you,

>R.

> > Ayush Sharma wrote:

> >

> > Hi Raymond

> >

> > By using the path attribute on RRs (only on Cisco)would force RR to send

> > all available paths instead of sending the best path. It is new feature

>and

> > cisco claims to make it available in early next year.

> >

> > Best Regards

> >

> > Ayush

> >

> > -----Original Message-----

> > From: Raymond Cheung

> > To: tpittman@frontrunner.eu.com; 'Peter J. Welcher'; mpls-ops@mplsrc.com

> > Sent: 20/12/2001 18:05

> > Subject: RE: MPLS load balancing

> >

> > Peter,

> > Tim just explained exactly what I'm doing. Without RR in MPLS network, I

> > could have multipath for BGP if there are multiple IGP path to a PE.

> > With

> > RR, I can't make multipath work. Someone suggests using different RD for

> > same VPN (same route-target), but I can't make it working. I want to see

> > anyone have done this before.

> > best regards,

> > Raymond

> >

> > -----Original Message-----

> > From: tim pittman [mailto:tpittman@frontrunner.eu.com]

> > Sent: Thursday, December 20, 2001 AM 11:10

> > To: 'Peter J. Welcher'; mpls-ops@mplsrc.com

> > Subject: RE: MPLS load balancing

> >

> > I think Rays issue is if a VPN destn. is actually reachable via two PE's

> > (PE1 and PE2).

> > eg a CE is dual homed to two PE's, PE1 and PE2.

> >

> > Then on PE3 because you can't iBGP load share via an RR, then only one

> > iBGP

> > route to the VPN destination (via either PE1 or PE2) is ever seen by

> > PE3.

> > The RR only passes on its best route.

> >

> > So doesn't matter that IGP on PE3 may then load share over differnet

> > internal paths

> > to reach either PE1 or PE2 , PE3 only ever sees one egress PE to reach

> > the

> > VPN destination.

> >

> > I guess iBGP load sharing without RR's would let PE3 see both PE1 and

> > PE2 as

> > equal

> > cost exit points and you could then use both, but don't think (with

> > cisco)

> > you can do it with RR's.

> >

> > tim

> >

> > -----Original Message-----

> > From: Peter J. Welcher [mailto:p.welcher@att.net]

> > Sent: 20 December 2001 13:25

> > To: mpls-ops@mplsrc.com

> > Subject: MPLS load balancing

> >

> > Hi, Ray. I may be missing what you're trying to do, but

> > I suspect from the CE-CE you're trying to load balance

> > for MPLS VPN's.

> >

> > If you're doing MPLS VPN, BGP / MBGP is only used for the

> > next hop (exit from the SP cloud) determination. If the next

> > hop is a loopback on the PE router, the IGP is used for path

> > to that loopback. The IGP should load balance equal cost routes,

> > and MPLS label paths will also do this.

> >

> > HTH

> >

> > Raymond Cheung wrote:

> > >

> > > Hello,

> > >

> > > Has anyone tried BGP load balancing in a MPLS network with route

> > reflectors?

> > > I have no problem to get the load balancing working without route

> > > reflectors. But since route reflector can only advertise one best path

> > for

> > a

> > > prefix, so I have to use different RD to make vpnv4 addresses look

> > different

> > > on route reflectors. The problem is, after using different RD (still

> > same

> > > route-targets), CE-CE reachability is broken.

> > > Please let me know if you have done this..

> > > Much appreciated.

> > >

> > > best regards,

> > >

> > > Raymond Cheung

> > > Senior Network Engineer (CCIE R&S)

> > > Schlumberger Network Solutions

> > > Tel: 713-513-3285

> > > Tel: 713-513-3000

> > >

> > ------------------------------------------------------------------

> > Dr. Peter J. Welcher Home Office: (410) 626-7735 or 7122

> > CCIE #1773, CCSI 94014 e-FAX: (208) 567-2310

> > Chesapeake Netcraftsmen, Inc. EMAIL: pjw@netcraftsmen.net

> > 1290 Bay Dale Drive, #312 ALT EMAIL: p.welcher@att.net

> > Arnold, MD 21012-2325 //home.att.net/~p.welcher

> > CELL PHONE: (443) 995-4859

> >

> > Chesapeake Netcraftsmen is a new consulting company dedicated to

> > quality. It is formed of ex-Chesapeake Computer Consultant and

> > Mentor Technologies employees with deep consulting and training

> > experience. We are currently 6 CCIE's. Our expertise includes

> > high-end routing and design, VoIP, QoS, MPLS, network management,

> > security, IP multicast, and other areas.

> > ------------------------------------------------------------------

> >

> > -------

> > 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

> >

> > -------

> > 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

The MPLS-OPS Archive

RE: MPLS load balancing