[rrg] [lisp] Fwd: [GROW] Operational experience with cachebasedmapping ID

[Scott Brim]

Excerpts from Tony Li on Wed, Nov 19, 2008 11:16:20PM -0600:
>  
> 
> |They can't be completely decoupled, because different data plane
> |mechanisms have different mapping requirements -- many none at all --
> |and (as you say) different mapping mechanisms work better with
> |different data plane mechanisms.  I believe you need to evaluate
> |data/control mechanism combinations.  
> 
> Fair, but from a modularity standpoint, we should see them as separate
> subsystems and attack them as such.
> 
> Tony

Yah.  Generate ideas about them as independent systems, but evaluate
them based on how they mesh.