Hi David,
> Jim,
>
> At 10:22 PM 4/30/2001 -0400, Jim Bound wrote:
> >With all due respect IPv6 is far superior to IPv4 for renumbering.
> >Have you looked in depth at Neighbor Discovery, Stateless
> >Autoconfiguration, and Router Renumbering RFCs. Then put them all
> >together.
>
> My understanding that all of these were in a significant state of flux with
> no significant deployment even possible at this time. I'm glad to hear my
> understanding was mistaken. Need more time to read...
No flux here. Router Renumbering is still being built. But the specs are
very solid.
> >Nothing I mean Nothing exists like this in IPv4.
>
> True. However, IPv4 does, at least, have deployed DHCP which seems to be
> of more interest to larger sites than stateless autoconf.
I do agree that Sateful Addrconf must be deployed and folks fought many of
early IPv6 people pretty hard on it. In DHC WG we hope DHCPv6 will be
wrapped up for London to move to PS status and that will get it done. As
you know there are implementors addressing that requirement now in the
industry and your on that list.
But this discussion was the abstract regarding Renumbering.
If your arguing IPv4 is OK and all is well. Hmmm. I see not point of
that argument and won't participate as it simply is not OK anymore.
> >We have never said we solved all the renumbering
problems. >
> I don't believe anyone has claimed the problems have been solved. They are
> difficult problems. I was under the impression, however, that A6 was
> considered a significant portion of the answer to the renumbering
> problem. If it is not considered such, then it should be dropped
> immediately as the additional complexity it imposes on the DNS is far too
> high to justify its deployment.
>
> >Nor did we say we would.
>
> For what it is worth (not much, I know), my personal belief is that trivial
> renumbering is key to IPv6's success. Without it, NAT will have a
> significant advantage, despite NAT's acknowledged problems (no intention of
> sparking yet another NAT flamefest, yes NAT is a blecherous hack).
I don't disagree that renumbering is an advantage. NAT cannot support 3
billion Wireless devices and except for the U.S. none have the address
space to support key end-to-end solutions. Thats why IPv6 will be
deployed. There are next steps to renumbering no one is saying there is
not. But how about a draft instead of just complaining from someone.
I just don't see the benefit of such a discusion for me anyway.
> >The reason IPv6 will be and has begun
> >deployment is because of "address space". Its the primary motivator.
>
> This is worrisome. By and large, I believe the main reason people are
> unhappy with the current situation with IPv4 is end user sites are normally
> rejected for address allocations of provider independent address space,
> forcing those sites to be dependent on their service provider for address
> space. The reason for rejection is typically _not_ a lack of address
> space, but rather the RIRs have been put into the position of trying to
> limit the number of provider independent prefixes which enter the routing
> system.
Same thing. THe RIRs are trying to manage a scare commodity.
> Given the current IPv6 addressing and routing architectures (as I
> understand them -- I'm sure I'll be "gently" corrected if I'm wrong), those
> same end user sites should get rejected for IPv6 space as well -- they
> should go to their service provider for their address space. Without
> trivial renumbering, those sites that transition to IPv6 will find
> themselves in _exactly_ the same place they are with IPv4, if not
> worse. They must renumber their site to their new provider's address
> space. All IPv6's bigger addresses would mean is that they have more to
> renumber. At that point, why fight the IPv4 inertia? Why not simply go
> with IPv4 NAT solutions, despite their warts (yes, I know what those warts
> are, but note that I can go down to Office Depot and buy a 56Kbps asynch
> NAT router off the shelf _today_)?
People are free to do that. We must be talking to different customers the
ones I talk to have had it with this present world and want IPv6 and not
NAT. But I don't just talk to U.S. Customers either :----)
But there are some very large U.S. customers that want IPv6 yesterday.
We should fix Multihoming. But that is not an addressing or routing issue
with IPv6 but something that was not fixed in IPv4 either. If we fix it
in IPv6 thats another advantage and I think we can.
What exactly in technical terms have you heard is wrong with IPv6
addresses and routing? Hard for me to defend inuendo......
> I believe A6 provides a useful generalization that can support greatly
> simplified renumbering. I also believe a significant amount of operational
> deployment experience is necessary to determine whether or not the
> complexity it imposes on the DNS is justified by the benefits it
> brings. At least at this point, there are implementations that will permit
> experimentation. Those experiments should be the focus of efforts, not
> attempts to derail the existence of A6 without supporting evidence.
>
I am not arguing about A6 anymore.
regards,
/jim
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
