Re: Draft on Globally Unique IPv6 Local Unicast Addresses

[Alain Durand <Alain.Durand@xxxxxxx>]

On Tuesday, May 13, 2003, at 10:58  AM, Pekka Savola wrote:

> On Tue, 13 May 2003, Bob Hinden wrote:
>>> Bob's proposal explicitly defines a type of addresses that are not
>>> expected to be routed throughout the Internet.  So, these addresses
>>> are local, not globally routable.
>>>
>>> However, Bob's proposal defines addresses that are unique throughout
>>> the Internet.  No single address identifies one node when used within
>>> one "scope" and another node when used in another "scope".  So, these
>>> are not "scoped addresses", as defined in the IPv6 Scoped Addressing
>>> Architecture.
>>>
>>> I *think* that Bob and I are using the same definition for the
>>> term "scoped", but I hope he'll correct me if I'm wrong.
>>
>> I tried hard to not define the term "scoped address".  I only used 
>> the word
>> "scope" once in the draft and that referred to global scope 
>> addresses.  I
>> may remove that use in the next version :-)
>>
>> The globally unique IPv6 local unicast addresses are not limited in 
>> their
>> use to any scope less than global by their design because they have
>> prefixes that are globally unique.  By comparison site-local 
>> addresses were
>> limited to a limited scope by their design because they had 
>> non-globally
>> unique prefixes.
>>
>> So yes, I think we agree.
>
> What would you propose to do with default address selection, if any?
>
> Should a preference in it be given to truly global or globally unique
> local addresses, or should these be indistinguishable from the point of
> view?
>
> This "addresses tried first" ordering is of some importance -- which 
> one
> reason for having this "scope" concept.

Pekka is right. The real question behind Bob's proposal is not so much
to know if it creates scopes or not, but what is the impact on the 
applications.

If applications need to be aware of those addresses and take different 
actions
if/when different combination of "bob's" or "regular" addresses exist, 
we
end up with much of the issues we had with Site Local, ambiguity 
excepted.
This needs to be clarified.

	- Alain.


--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to majordomo@sunroof.eng.sun.com
--------------------------------------------------------------------