[prev in list] [next in list] [prev in thread] [next in thread] 

List:       ipng
Subject:    RE: [dhcwg] Review of draft "Prefix Assignment in DHCPv6"
From:       "STARK, BARBARA H" <bs7652 () ATT ! COM>
Date:       2012-12-13 19:26:40
Message-ID: 2D09D61DDFA73D4C884805CC7865E611302168ED () GAALPA1MSGUSR9L ! ITServices ! sbc ! com
[Download RAW message or body]

> Just I mentioned in previous email, SLAAC is optional WiMAX deployment.

The attempt to create an access network without RA/RS is nothing new.
Other (e.g., DSL, PON) access network technologies have considered this and \
determined that the biggest missing piece is route info. Which is the reason behind \
draft-ietf-mif-dhcpv6-route-option. But on point-to-point links it is possible to \
assume the next hop, so that route info isn't needed. A solution without RA is simple \
on point-to-point links.

I believe the existing address and prefix DHCPv6 options are sufficient to support \
what you seem to be trying to do. Unless the goal is to be able to share a single \
prefix among multiple WiMAX endpoints? I'm not familiar with prefix sharing across \
point-to-point connections, so I'm guessing the expectation is one /64 prefix per \
WiMAX endpoint?  
In which case, it would appear that a host could request a /64 IA_PD, and then use \
that prefix for whatever it wanted. The host doesn't have to be a router, or offer \
the prefix to any other host. It can create any addresses it wants for itself. In the \
CE Router (RFC 6204) requirements, the "unnumbered" model is already described, where \
the RA contains no "A" prefix and no IA_NA is offered; so it takes addresses for \
itself from the IA_PD. Your proposal seems to be exactly this, with the only \
difference being that the CE router does try to also offer the IA_PD prefix on its \
LAN. But since there's nothing policing this (the access network has no clue what use \
the IA_PD is put to), there's nothing to prevent the host from just keeping the \
entire IA_PD for itself. Why does this not meet your need?

For a WiMAX-defined endpoint using IA_PD would be no more difficult that what you are \
proposing. In fact, it would be pretty much the same. For the DHCPv6 server it would \
be simpler, because the server would not need to have different options that function \
identically (from the server's perspective). The DHCPv6 server truly does not care \
what use the IA_PD prefix is put to, or whether the requesting DHCPv6 client is \
really a "requesting router" or just a device who wants a prefix all to itself. \
                Barbara
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------


[prev in list] [next in list] [prev in thread] [next in thread]