Re: [mpls] working group last call on draft-ietf-mpls-ldp-igp-sync-bcast-00.txt

[Mikael Abrahamsson <[e-mail]>]

On Wed, 3 Mar 2010, Sriganesh Kini wrote:

> [Sri] If both ends are not capable of LDP/IGP Sync it seems like a 
> misconfig (or lack of a config knob)

Correct, but still valid (would mean LDP/IGP sync would work even if only 
one end of the link had it implemented.

> [Sri] If the goal is to achieve overload-bit like behavior with the 
> exception of allowing transit traffic through the router when no other 
> path is available, then it can be achieved with "max-metric 
> all-interfaces" on one router even though this results in each link 
> being uni-directionally at max-cost. This is because during SPF 
> computation, to transit the router, it will have to go through one link 
> in the direction that it is advertised with max-cost.

That doesn't solve directly connected interfaces in HSRP/VRRP type 
redundancy scenarios.

-- 
Mikael Abrahamsson    email: [e-mail]