On Sun, 24 Aug 2003, Tony Hain wrote:
> > This document seems to take for granted that local-use addressing is
> > needed.  Moreover, it lists requirements for every possible case where
> > local-use address could be applied to (and, not for example, those
> > cases where the local-use addressing is really necessary).
> 
> Necessity is both the perception of the network manager in trying to
> implement a local policy, and the availability of alternatives that
> actually fit all of the local constraints.

It may not be our business to enhance misguided perceptions of the network 
managers, rather than correct them.

> > Process questions:
> > 
> >  1. Shouldn't we first see the requirements for site-local 
> > replacement (and other issues) and not jump straight to the 
> > requirements for local addressing?
> 
> I don't understand the question. My original draft started from the
> premise that the WG should at least have the requirements for local
> addressing before deciding that any given technology is either
> acceptable or unacceptable. Site-local is a specific technology that
> meets many of the needs of local addressing, but creates other problems
> through its ambiguity. Getting rid of ambiguity does not remove the
> requirements for local addressing.

The ambiguity was just one reason to get rid of site-locals, and I believe 
a non-ambiguous version is already being worked at.

However, my point is that we should not be advocating local-use addresses.  
It is far from clear where they're actually needed, and where they're 
actually the best solution.

I.e., I think there was some sort of consensus in the SF IETF meeting that 
there is a need to do at least *something* after we deprecate site-local 
addressing; there are some scenarios where SL's were being imagined to be 
used -- and now we should figure out how to address the needs of those 
scenarios *in general terms*, rather than jumping straight to the 
requirements for local addressing.  It might just be that we don't need 
local addressing in all of those scenarios (actually, I'm certain of 
that).

> >  2. Then if we see that local addressing is required to do X, 
> > see that 
> > this document covers X adequately?
> 
> If the goal is to design specific approaches for every scenario, then
> this suggestion would be appropriate.  The goal of the requirements
> document at the moment is to aggregate multiple needs under a single
> technical approach. If the WG decides that multiple approaches make more
> sense, then there will need to be multiple requirements documents.

Agree.

Note that I'm not really strongly advocating separate documents, but that
we AT LEAST figure out why such-and-such requirements are there.

> > I don't send in more detailed comments on the draft, because 
> > I pretty much 
> > disagree with everything it says.  
> 
> I understand your network does not have these requirements, and if we
> required everyone to implement local addresses I could understand some
> degree of concern. What I don't understand is why you believe it is
> appropriate to tell another network manager that his stated requirements
> are invalid.

I do not want a cure that is worse than the disease.  That is, if we make
local-use addresses work "too well", and applicable also in scenarios
where it isn't needed -- people go for locals and not globals.  That would
be a serious disadvantage for IPv6.
 
> >  Just two notes:
> > 
> >  3.1 -- "Network managers have stated, and historical experience has 
> > shown, that there is a need for addresses that do not require public 
> > registration."  
> >  ==> there is no supporting evidence of this expect vague 
> > statements.  
> > Please be more explicit as I don't see how we can take this for given.
> 
> Chasing down quotes is not appropriate. If you want to see more text about
> people grabbing random space from documentation or currently using 1./8, I
> suspect we can come up with that.

I fail to see how this leads to the requirement for 
non-publicly-registered addresses.  What's the problem with  public 
registration?  Perhaps there is just a problem how that has been done with 
IPv4.

[snip]

-- 
Pekka Savola                 "You each name yourselves king, yet the
Netcore Oy                    kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings



--------------------------------------------------------------------
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 [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to