[c-nsp] BFD w/ static routes

[Rodney Dunn]

Well, you could use IPSLA with EOT (enhanced object tracking)
and tie the static routes to the object on the CE side.
That would reroute the CE static route to the backup path
when the object goes down.

You would have to do the same thing on the hub for
the object pointing towards the spoke to remove
the hub static route when there is a loss of
connectivity to the spoke so you don't have a unidirectional
packet forwarding problem.

I've never done it but it looks like IPSLA probes may be
vrf aware so you could put it on the hub.

http://www.cisco.com/en/US/tech/tk648/tk362/technologies_white_paper0900aecd8017f8c9.shtml

I just tried to configure one in the lab on 12.4(1.5) code I
happened to have there:

ip sla monitor 1
 type udpEcho dest-ipaddr 1.1.1.1 dest-port 20
 vrf test

so I'd assume the probe is VRF aware.

This should serve as an interim solution although I strongly
agree that a simple BFD solution on the link tied to the
static routes directly or indirectly through a tracked object
would be much easier to implement. I'm working with development
on trying to move that forward already.

Rodney



On Thu, Aug 04, 2005 at 05:08:00PM +0300, Pekka Savola wrote:
> Hi,
> 
> On Thu, 4 Aug 2005, Rodney Dunn wrote:
> > CSCsb48249
> > Externally found enhancement defect: New (N)
> > Request for BFD integration with object tracking
> >
> > It's on the roadmap from what I've been told.
> 
> Thanks!
> 
> > Could you explain the exact topology you want this
> > for?
> 
>                 I primary link
> --- [PE rtr] ----- customer1 rtr
>         |       I         |
>         |       I         |
> --- [PE rtr2] ---- customer1 rtr2
>                 I standby, backup link
> 
> The customer has two links to the ISP.  One is never used except for 
> backup purposs (control messages to ensure liveness is OK, but no 
> load-balancing is needed in our case, though that's possible as well).
> 
> The ISPs PE routers run iBGP or an IGP between each other.  The 
> customer is routed with static routing.  When the primary link goes 
> down, due to media converters or whatever, the PE and customer router 
> may still think the link is up.  The customer's prefix is still routed 
> towards the failed link (and the same for customer's default route 
> pointing to the ISP) and traffic is blackholed.
> 
> This problem is typically solved by running BGP or an IGP to detect 
> link failures and reroute.  We don't want to run an IGP across an 
> admin domain, and setting up and maintaining BGP is too complex for 
> the customer.
> 
> Hence, the solution is to use BFD to check the liveness of the link 
> and static routing.  When the link loss is (reliably) detected, the 
> static routes become inactive, and are removed from PE rtr <-> rtr2 
> iBGP/IGP, and the lower-priority static routes at backup routers get 
> active.
> 
> Hope this clarifies.  If there are other ways to solve this problem 
> (apart from using routing protocols or BFD), pointers are welcome..
> 
> -- 
> Pekka Savola                 "You each name yourselves king, yet the
> Netcore Oy                    kingdom bleeds."
> Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings