Re: [Bier] DISCUSS? SI vs. sub-domains - draft-ietf-bier-architecture-02

[Caitlin Bestler <[e-mail]>]

I meant to type "adjacencies". Apologies for not catching the 
auto-correction.

Anyway, from your explanation I think I have a fairly good understanding.

The key *intended* difference is that the way sub-domains are *intended* 
to be
used the end nodes would be automatically assigned to the sub-domains based
upon the routing underlay.

But the architecture document makes it clear that there *may* be 
multiple sub-domains
using the same routing underlay, in which case the assignment of end 
nodes to sub-domains
starts to look a lot like just assigning a BFR-ID.

Is there some succinct explanation as to why a deployment SHOULD NOT map 
multiple sub-domains
to a single routing underlay, but they MAY do so for the following 
reasons... ?

If sub-domains SHOULD map to a unique routing underlay while SI-subsets 
MUST map to the same
routing underlay then you have a justifiable distinction. But 
sub-domains that MAY map to the same
or different routing underlays while SI-Subsets MUST map to the same 
routing underlay as the SI just
seems like excessive detail in the API.