Re: I-D Action: draft-ietf-6man-rfc6434-bis-01.txt

[Brian E Carpenter <[e-mail]>]

On 21/09/2017 06:42, Manfredi, Albert E wrote:
> -----Original Message-----
> From: ipv6 [mailto:[e-mail]] On Behalf Of Templin, Fred L
> 
>> I hear you, and I am not suggesting any change in the host and
>> router definitions. However, an end system that acts as both a
>> router and a host has to do something slightly different than
>> an ordinary router would.
> 
> This is the typical case for any "dual-homed host," in my experience. An end system connected redundantly, for reliability/survivability reasons. They are routers, but configured to not forward packets and not announce routes (e.g. using RIP or OSPF). Very common in some environments, and frequently problematic.
> 
>> Whether/not you agree that the term "end system" should be
>> used to describe these, I still see value in importing the
>> weak/strong ES discussion from RFC1122 because the same
>> concepts apply to IPv6.

Not exactly. We discuss this a bit in RFC 8028, specifically
https://tools.ietf.org/html/rfc8028#section-1.1 , and there
are some subtle differences.

Personally I would prefer that we don't add dependencies on IPv4
documents. We might want to make all of those Historic in a few years.
 
> I think this issue also relates to what an IP address refers to. I'm under the impression that in IPv6, we insist this must be the address of an interface, rather than a host. In IPv4, it's more ambiguous, which is also problematic (IMO). If we insist that IPv6 addresses belong to interfaces, then the weak model would not apply to IPv6, no?

Again, see RFC 8028.
 
> 
> (This is also similar to MAC addresses, whether they should apply to a host or to an interface. A debate from a previous era.)

But it doesn't really matter for IPv6. As long as the MAC address is unique
on each link, everything works.

    Brian