[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Questions/Comments (was: RE: Accept hain/templin draft as wg item?)


I am quite satisfied with the coverage of the requirements for local
addressing. That said, I have a few questions, and comments.


1) What does the 3rd sentence in the 2nd para in section 3.4 mean? The
sentence I am referring to is "Given the presence of the well-known prefix, an
application that chooses to check can infer that there is an explicit....".
What does the application choose to check in this case? 

2) Section 3.5 mentions that absolute uniqueness is not required, but the
probability of collisions should be extremely low. We should add text to make
it mandatory for proposals to define following formulae:
	1) Formula for probability of collision when "x" networks are merged,
when "n" addresses have been chosen from the address space of "m" values.

	2) Formula for probability to choose a duplicate address after "a"
addresses have been chosen from the address space of "m". 

Presence of the formulae will allow comparing various proposals on the
uniqueness front.

Maybe we can also describe the values of x, a, and n (as proportions of m) in
this draft. I am not sure how to do it though.

3) The requirement for conflict resolution (section 3.5) through a central
registry seems to clash with section 3.1. On further thinking about it, I feel
we should change the wording of section 3.1, and not make the requirement for
not having public registration mechanism mandatory. I can't think of anything

4) I did not understand the requirement in section 3.6? My understanding is
that this section refers to both - PI and non-globally routed, and PI and
globally routed addresses. If this is so, can we have two subsections, and
state that PI, and globally routed is a advantageous, but a "MAY" requirement,
and if it is not satisfied, PI, and non-globally routed is "MUST".

5) I am not able to form the requirement stated in section 3.10. Can you
rephrase? In the 2nd para of section 3.10, what policies do you imply? Where
are these policies located? It also states that the policy decision moves from
the edge of the originating device. Where does it move to? I guess, I am
confused about the meaning of "policy". Once that is clear, I will get an
answer for the other questions.

6) In the summary section, modify the line about Address to say that the
"Address selection policy tables might need modifications to enable the
selection limited range address space over global addresses". Actual details
about the modifications should be specified in the proposals. 

7) In the same section, I don't quite understand the meaning of the word
"policy" in the first sentence of the 3rd para. 

   Looking a bit closely it seems that there is overuse the word policy. The
document talks of routing policy, operational policy, filtering policy,
one-size-fits-all policy, policy without any qualification in front. At a
minimum, all the usages of the word policy should have some qualification in

7) A high-level comment - Maybe we should add a section that says what
information should be present in the proposals for local addresses apart from
the requirements stated in Section 3. "Probability formulae" and "Effects on
Address Selection" can go there.

8) The draft uses both the terms "limited range" and "local use". Can we
narrow down to one term as much as possible? I prefer "limited range"


1. Add a section named "MANETS", and you can put all the examples of the
MANETS (section 4.3, and section 4.4) as subsections within that section.

2. Section 4.1 can be replaced by three sections - "Border filtering",
"Maintaining confidentiality of the address space", and "Test networks".

3. Move some of the examples defined in section 3.5 to the scenarios, and
explain them in more detail. Few on top of my head are - 

	- "Home networks", 

 	- A scenario concerning applications that use IP addresses in the ACL
or 	  configuration can be added. Such applications include license
servers 	  that use the IP address, firewalls within the site, web site
access 	  	  mechanisms to allow access to only certain subnets. Since
renumbering 	  is very difficult for these cases, it is a strong case for
stable IP 		  addresses.

4. Definitions/acronyms - PI, PA, VPN, limited range

5. Move the scenario in Section 4. to a subsection (may be a subsection under
the "Home Networks" section) like all the other scenarios. 

> -----Original Message-----
> From: owner-ipng@sunroof.eng.sun.com [mailto:owner-ipng@sunroof.eng.sun.com]
> On Behalf Of Fred Templin
> Folks - do we have consensus to accept this document as an
> IPv6 wg item (see below)?

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