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

Re: Draft on Globally Unique IPv6 Local Unicast Addresses



Tony,

You are hitting exactly on the points that concern me, and I
believe some implications for a local-use addressing scheme
become evident from what you've said. I think this is approx.
90% about the phiolosophical argument about L2 vs. L3; issues
with constructing networks out of differing L2 media types
can be overcome.

Your final question on how large does a network need to be
to 'require' rather than 'want' hierarchy is very poignant.
Seems like it might warrant some examination in the document.

Fred
ftemplin@iprg.nokia.com

Tony Hain wrote:
> Fred L. Templin wrote:
> 
>>I read the document, and it seems to me there needs to be a 
>>discussion of cases in which link-local addressing (fe80::) 
>>is insufficient. In my opinion, it comes down to identifying 
>>use cases in which flat vs. hierarchical local addressing 
>>schemes are most appropriate. I believe this is a very 
>>difficult subject to get one's arms around, but one that a 
>>local scope address requirements document needs to tackle.
> 
> 
> I guess I accept the premise that the document should deal with it, but
> it is not clear to me what form that would take beyond the direct
> evidence that people have deployed networks that are larger scope than a
> single link, and smaller scope than global. Their requirement for larger
> than a single L2 link is ??? 
> 
> Clearly compartmentalization and traffic isolation comes to mind, but
> how much of that requirement is based on anecdotal requirements from
> broken implementations of > 15 years ago. Routing convergence is another
> frequent reason, but again global flat L2 host routing is not the
> problem it once was. 
> 
> Is it sufficient to say that network managers insist on the protection
> offered by a scope less than global, but simply 'want' the
> organizational structure of an L3 hierarchy that is larger than a single
> L2 link? If not, what is the real requirement for hierarchy? Is this the
> philosophical argument about L2 vs. L3 being better, or are we really
> talking about something like the need for a network to be constructed
> out of differing L2 media types for cost reasons. If different media is
> the reason, the arguments are more about interconnecting dissimilar
> media than hierarchy. Which brings us back to the question of how large
> does a network need to be to 'require' hierarchy rather than simply
> 'want' it? 
> 
> 
> Tony
> 
> 
> --------------------------------------------------------------------
> 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
> --------------------------------------------------------------------


--------------------------------------------------------------------
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
--------------------------------------------------------------------