[homenet] Let's make in-home ULA presence a MUST !?

[Pierre Pfister <[e-mail]>]

Hello group,

The architecture document states the following:
- A home network running IPv6 should deploy ULAs alongside its globally unique prefix(es) to allow stable communication between devices [...]

This translates into section 9.1 in the Prefix Assignment draft:
- A router MAY spontaneously generate a ULA delegated prefix [...]

So, that MAY should probably be a SHOULD. But the reason for this mail is one level higher:

Question: Should the generation of a stable ULA prefix be a MUST in any case ?

Advantages would be:
- In the protocol design process, we could assume in-home IPv6 connectivity. No need for special case for IPv4-only connectivity, no need for special TLVs, flags, or whatever.
- In the implementation process, it is way easier to handle one single IP version for all in-home traffic. Let it be IPv6 !
- This connectivity would be more stable than IPv4 (which only exists when there is an IPv4 uplink). IPv6 enabled apps would therefore behave *better* than IPv4-only. Which would in the end help for transitioning.

Disadvantages are:
- The best (and probably only correct) way of advertising in-home ULA connectivity is using RIOs. Which some (e.g. apple's) devices don't support (yet?).
- Some currently existing implementation may fail when facing ULA vs IPv4 choice.


It looks to me that disadvantages will be overcome in the coming years if IETF requires implementation to handle ULAs and RIOs better.

So the question is, should I change the Prefix Assignment draft and make ULA existence a MUST so that we/developers can rely on that in our protocol design and implementation process ?

Cheers,

Pierre