Re: draft-ops-endpoint-mib-04.txt

[Dave Thaler <dthaler@dthaler.microsoft.com>]

Juergen writes:
> How do other people on this list think about this issue?

See my opinions below...

> Keith> Given the above, I now think that InetAddressType should be
> Keith> deleted from this MIB, and the AddressFamilyNumbers TC from
> Keith> IANA-ADDRESS-FAMILY-NUMBERS-MIB should be IMPORTed and used
> Keith> instead, This will promote greater consistency across IETF
> Keith> MIBs.

I agree with Keith.  (I was one of the people who originally
suggested this.)

> Some questions:
> 
> Where do you define the TCs for the various address families? Do they
> go under IANA control as well? Or do they go into RFC(s)? 

My preference: RFCs.

> If they go
> into RFC(s) and need additions over time, do we put them in multiple
> modules do avoid locks on the standards track? Or do we have no locks
> since the MIB modules only import InetAddress and not the specific
> TCs?

Don't care.

> How do we solve the problem with non-global IPv6 addresses? Do we use
> a 20 byte IPv6 address all the time (setting the scope identifier to
> zero most of the time)? We currently have two IPv6 address formats to
> work around this problem.

My preference: use a single format for all IPv6 addresses.
(The "scope indentifier" already appears to be overloaded with
different meanings for link-local and site-local anyway.)

> Is it likely that we can make IANA accept DNS names as an address
> family? (Or do you not care since you dislike the concept of DNS names
> anyway?)

My preference: it would be beneficial to make IANA accept DNS names
as an address family.  The HostAddressType TC in the REMOPS MIBs
(PING MIB, Traceroute MIB, etc) is another example of a TC that 
duplicates InetAddressType and requires DNS names.

Dave Thaler