I agree that connecting non-DHCPv6 hosts directly to the proposed architecture is problematic. It's good to ask about alternatives to solve the problem, to understand how big the problem is and to understand whether the alternatives are effective solutions.
I disagree that it's always better to provide options; sometimes, depending on the amount of work required to generate the options and the size of the problem being solved, it may be better to ask that the deployment conform to the existing options rather than defining new options. In this particular case, we need to balance the work required to define the RS-tagging option - allowing for future work which will likely be required to complete the work based on our experience with the similar extensions to DHCP - against suggesting that deployments adapt to the existing mechanisms which are designed to support the underlying network architecture (as I understand it). I understand that you are looking to accommodate subscribers that have devices that are not DHCPv6-capable and that are directly connected to the SP network. How many of those devices will actually be able to move to IPv6 and do we want to expend the effort in design and implementation to accommodate those devices, when there is the alternative of connecting those devices through a home gateway? - Ralph On Sep 9, 2010, at 3:17 PM 9/9/10, Alan Kavanagh wrote: > Hi Ralph > > I look at this where we have hosts that are SLAAC based only as Suresh has > noted and are directly connected to the Edge Router. In this case we have > hosts that connect via SLAAC as is typically the case for unmanaged devices. > In the case of these hosts behind a Routed RG that do SLAAC and not DHCPv6. > In the case where these hosts are bridged through the RG either selectively > or the RG is L2 they still use SLAAC for address configuration. > > I think we need to ensure that we don't force Service Providers down a path > that they are reducing the flexibility in how they should deploy their > network and services. What we need to do is ensure we have options for the SP > to choose the most appropriate solution that fits their network deployment > and business needs. > > Im not here to compare SLAAC vs DHCP, that's not what the draft does and > that's not the intention.....as I see both protocols existing and having > merits for different deployment needs......its not my place to tell O.S > vendors, CPE vendors, SP's and home users how best to suit their needs, its > better to provide options, and embrace that we have SLAAC and will be used in > numerous deployment cases. > > I don't see these as corner cases Ralph there are a significant number of > SP's/subscribers whom connect to DSL networks in this manner and as we want > these SP's to embrace IPv6 we need to remove as many "roadblocks" as possible. > > Alan > > -----Original Message----- > From: [email protected] [mailto:[email protected]] On Behalf Of Ralph > Droms > Sent: September-08-10 2:27 PM > To: Joel M. Halpern > Cc: Narten Thomas; IPv6 WG Mailing List; Suresh Krishnan > Subject: Re: New version available (Was Re: Consensus call on > adopting:draft-krishnan-6man-rs-mark-06.txt) > > Joel - only some operators have decided that they need to allow for the > corner case of an IPv6-capable device with no DHCPv6 connected directly to > the SP network. CableLabs took the approach of mandating DHCPv6 for any > device connected to a cable SP network; the expectation being that a high > percentage of users use some kind of home gateway, and those gateways will > all include DHCPv6. Inside the home, the home gateway provides RAs > configured for the hosts to use SLAAC. > > I object to expending IETF energy on extensions to the ND protocol > architecture to solve a corner case for which DHCPv6 exists as the designed > solution. > > - Ralph > > On Sep 8, 2010, at 8:01 PM 9/8/10, Joel M. Halpern wrote: > >> Doug, I am confused by your comments. >> >> Let me describe how I understand the situation. We claimed, when we crafted >> IPv6, that hosts did not need to use DHCP for address assignment. As such, >> many host stacks did not use DHCP for address assignment. >> >> Now, operators wanted to offer IPv6 service. I hope we think that is a good >> thing. For residential, they looked at what they could count on from the >> hosts. And some of them concluded that they could not count on DHCP, so >> they designed an architecture around SLAAC. In other words, they ddi what >> we told them to do. >> >> There are other constraints, and the problems that result are not simple to >> solve. But they do not matter for your comment. >> You seem to be saying that even though the operators did what we told them, >> we should tell them "sorry, you need to redesign your network, and you need >> to assume DHCP based address assignment for all devices in the home, even if >> you can not count on actually having that work??" >> >> Yours, >> Joel >> >> >> On 9/8/2010 1:29 PM, Doug Barton wrote: >>> Ralph, >>> >>> I use IPv6 in XP so I can confirm your suspicion on both counts. >>> IPv6-only is a non-starter for XP. >>> >>> Suresh, >>> >>> I understand your goals quite well, which is why I'm opposed to the >>> adoption of the draft. :) Since practically Day 1 of the IPv6 effort >>> there has been a movement to make SLAAC/RA a full-featured >>> replacement for DHCP, by hook or crook. Please note, I am not >>> impugning _your_ motives, but I am opposed to your goal. >>> >>> >>> Doug >>> >>> >>> On 09/08/2010 09:17 AM, Ralph Droms wrote: >>>> Nit: seems unlikely to me you will have any XP devices running >>>> IPv6-only; if my understanding of the situation is correct, such a >>>> device requires manual installation of the IPv6 stack and still >>>> requires IPv4 for DNS. >>>> >>>> - Ralph >>>> >>>> On Sep 8, 2010, at 5:36 PM 9/8/10, Suresh Krishnan wrote: >>>> >>>>> Hi Doug, >>>>> >>>>> On 10-09-08 02:02 AM, Doug Barton wrote: >>>>>> On 09/07/2010 06:38 AM, JOSHI, SHRINIVAS ASHOK (SHRINIVAS ASHOK) wrote: >>>>>>> 5. Creating an alternative to DHCPv6 ? >>>>>>> >>>>>>> One SLAAC is defined to do functionality similar to DHCP >>>>>>> (including per host prefixes/options) how long before options are >>>>>>> added so SLAAC becomes an alternative to DHCPv6 ? >>>>>> This is the basis of my opposition to adopting the draft >>>>>> (expressed neatly here, as well as by other authors in this thread). >>>>> >>>>> As I said in my response to Shree, the goal is to provide support >>>>> for SLAAC-only IPv6 clients. It is not a practical option to >>>>> require support for stateful DHCPv6 for clients that are no longer >>>>> supported (e.g. XP). I agree that going forward, stateful DHCPv6 >>>>> will be a solution for this problem. >>>>> >>>>> Thanks >>>>> Suresh >>> -------------------------------------------------------------------- >>> IETF IPv6 working group mailing list >>> [email protected] >>> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 >>> -------------------------------------------------------------------- >>> > > -------------------------------------------------------------------- > IETF IPv6 working group mailing list > [email protected] > Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 > -------------------------------------------------------------------- -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
