Re: Uneasy about metalevel=1 at the subscriber boundary

[Paul Francis--formerly Tsuchiya <[e-mail]>]

>  
>  I am in the process of reading the Address Concentions draft that was
>  announced today.  I am somewhat uneasy about the assigment of
>  metalevel=1 and higher to be internal to subscribers.  It seems to me,
>  knowing how the topology looks in my general area, that it would make
>  more sense to put this boundary at metalevel=2.  The same problem
>  which you attempt to solve at the subscriber boundary could also be a
>  problem at the local-access-provider boundary.  Consider the case of
>  NEARnet, which---last I checked---was attached to NSFnet, AlterNet,
>  ESnet, TWBnet, and probably several others that aren't listed in the
>  map that I have.  This means a potential level-numbering conflict for
>  NEARnet (=== `LAP') relative to BP[1-4+].

But, my intent is to give even LAPs (which I'm not sure I would classify
NEARnet as) top level assignments.  So, the the numbering tree, they
are at the top.  If it is not desirable to advertise NEARnet globally,
then you would form an address using a route fragmant (a "horizontal"
relator, that is....NSFnet-H-NEARnet-V-Subscriber....

However, this point doesn't really change your argument, anyway.
So, perhaps we should put the provider/subscriber metalevel at
metalevel=2, thus allowing one metalevel above and one below....

PX