On Thu, Feb 22, 2001 at 09:51:49AM -0600, Scott.Keoseyan@BroadWing.com wrote:
> Hello,
> 
> I am trying to implement traffic engineering tunnels in a lab environment to
> force traffic from specific VPNs across specific links, without any success.
> Does anyone have any examples of how this is done?  
> 
> I can engineer traffic for destinations in the main routing table but not
> for any destinations in any of the VRFs.  When I try to build a TE-Tunnel
> for a destination in a VRF, it simply does not come up.  The path signalling
> using RSVP seems to also work fine for destinations in the main routing
> table but will not work for any destinations in any of the VRFs.  
> 
> I guess what I am asking is, how do I build a TE tunnel for a particular
> destination associated with a VPN and force that VPN to use a strict path
> first and then failover to another?  
Since you can't build TE tunnels to a particular VPN, there's a few
steps you can do to work around this:
1) put multiple loopbacks on the egress PE (loop1, loop2, etc)
        loop1 = 1.1.1.1
        loop2 = 2.2.2.2
        etc
2) set the next-hop for different VPNs to different addresses on that
egress PE
        vpn foo next-hop = 1.1.1.1
        vpn bar next-hop = 2.2.2.2.
3) build TE tunnels to the RSVP RID on the egress PE (i.e. loopback0)
        tunnel destination x.x.x.x (*not* loop1 or loop2)
4) don't run autoroute on the TE tunnels, but point static routes for
the next-hops down these tunnels
        ip route 1.1.1.1 tun1
        ip route 2.2.2.2 tun2
        etc
Note that step 1 isn't actually necessary, but I find it helps to save 
on confusion.  
eric
> 
> Thanks,
> 
> Scott
> --
>  Scott A. Keoseyan (sak@broadwing.com)
>  Principal Engineer - Lab Services
>  B R O A D W I N G    Inc.*
>  1881 Campus Commons, Suite 210
>  Reston, Virginia 20191
>  (703)391-1831 - (FAX)391-1810 
>  http://www.broadwing.com/ccielab
> 
> 
> * Company name mentioned for identification purposes only.  
>   These ramblings are my own opinions
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:30 EDT