Re: [narten@xxxxxxxxxx: PI addressing in IPv6 advances in ARIN]

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

 



At 15:13 -0400 4/14/06, Noel Chiappa wrote:

There's a certain deep irony here, because PI-addresses have been considered
at length in the IETF in at least two different WG's - CIDR-D and Multi-6.
Both rejected them after extensive discussion.

Can you point to documents that give the results of this? (Please don't cite mail archives, following threads is not very efficient.) If the reasons for the rejection are obscure, the uninformed will be likely to stay that way.

Nevertheless, a policy-making body has seen fit to ignore that, and make an
engineering decision to deploy PI-space. It's hard to read that decision any
other way than to have it imply that the decisions in those WG's were
technically uninformed.

It's hardly an engineering decision. The sense was that without PI space, IPv6 will remain in a state that will not get it deployment experience. Having (personally) sat through the discussions at ARIN, RIPE, and APNIC, and having seen IPv6 grown in the IETF, I wouldn't characterize the anything as "uninformed."

The debate between 1) what happens to router tables with PI prefixes, 2) what happens when the protocol is crimped into a comfortable-to-operate fashion, and 3) what customers are will to bear, has been raging for more than a year.

It's plain to see that PI space threatens to blow up routing.
It's plain to see that IPv6 is supposed to allow for end-to-end flexibility.
It's plain to see that customers no longer want to be tied to ISPs.

My personal take is that IPv6 answers to a set of requirements devised 10 years ago and that no longer captures all the needs of the Internet. Route aggregation is good, but it relies on business assumptions that are unrealistic. Further, layer violations have become rampant, even if they are architecturally distasteful, they have become part of the operations landscape.

In today's Internet, an "end user" whose cash flow depends on the network will be willing to rely on one provider, nor be willing to be locked into any set of providers over time. (Note that providers are the least financially stable players in the game.) Even if you have two providers today - what if they merge? I've known many who thought they had redundant paths, only to see that through mergers on the other side, all they had was on the same fiber.

Many times customers walk because of business decisions, as in disputes over billing practices. This was added to a comment I had made within ARIN some time back. The moral is that an assumption that address assignments/allocations from RIR to LIR to subLIR to user isn't universally loved.

Renumbering is not pleasant. Renumbering IPv6 style (prefix from the infrastructure, suffix from something else) isn't as promised thanks to things like firewall rules and other in-network beasts that configure addresses. You can say that these middle boxes don't belong, but try running an enterprise network without them. (I haven't been without a firewalled and/or NATted address since 1996, even through different employers.)

Before IETF'ers start pointing fingers at "uniformed" decisions of other groups, it's worth asking ourselves where the misinformation comes from and, possibly, whether what has been developed is still relevant. I hear folks saying the IPv6 has bigger addresses and there's no other option, so we might as well go to IPv6. Is that all it comes to, 96 more bits than v4 and the only game in town? (I ask rhetorically.)

--
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Edward Lewis                                                +1-571-434-5468
NeuStar

Nothin' more exciting than going to the printer to watch the toner drain...

_______________________________________________

Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf

[Index of Archives]     [IETF Annoucements]     [IETF]     [IP Storage]     [Yosemite News]     [Linux SCTP]     [Linux Newbies]     [Fedora Users]