[Idr] BGP Layer 2 Encapsulation Types

[tom petch <[e-mail]>]

The IDR WG maintains an IANA registry for BGP Layer 2 Encapsulation Types which the OPSAWG WG is now turning into an IANA-maintained YANG module but whereas the registry entries use values such as 
17  Structure-agnostic E1 over packet      
18  Structure-agnostic T1 (DS1) over packet 
43 DS1 (ESF) Nx64kbit/s with CAS using    Structure-aware     
the module entries use text strings such as
identity e1 
identity t1 
identity cas 
with no mapping between the two.   As and when a new encapsulation is defined, IANA is expected to add an entry to the YANG module  where 'The name of the
   "identity" is a lower-case version of the encapsulation name provided
   in the description.' which is clearly not the case for many of the initial entries, and I have commented on this at Last Call.

I then wondered how easy it will be for users to know the mapping between registry and module, which, as I say, is not specified.  I did think that the IDR WG could add a column to the registry to register a suitable text string, I then thought that it should be the IDR WG that chooses suitable values but finally wondered why OPSAWG are using identity at all, as opposed to a YANG enumeration, which would be much closer to the BGP registry and have the mapping built in, with a string and value both being part of the definition.

There are occasions when an identity is preferable to an enumeration - an identity can be extended by anyone in any direction which can be a benefit - but since change control of the YANG module is by IANA here, that consideration does not apply.

Tom Petch