Re: New Version Notification for draft-koodli-ipv6-in-mobile-networks-02

[Rajeev Koodli <rkoodli@cisco.com>]

On 4/19/10 4:07 AM, "teemu.savolainen@nokia.com"
<teemu.savolainen@nokia.com> wrote:

> Yes, that meeting and discussion before that. The host based transition
> mechanisms were rejected because transition was agreed to be more easily done
> by networks than by all hosts - and hosts were discussed to have dual-stack
> connectivity or IPv6-only connectivity. I agree these discussions sound
> echoing, they are, but for me it seems there is interest to restart this
> discussion again as the draft that, from my point of view, talks about host
> based transition mechanism called "DHCPv4 on-demand", was adopted...

Let me assure you that I am *not* interested in restarting a new discussion
on transition mechanisms, that too a *host-based* one! :-)


> 
> In my opinion the draft can mention it is possible to build such IPv4 address
> on demand system, but for it to be really useful it would have to be supported
> by wide variety of hosts, and that is not at all certain because as said many
> times, DHCP is optional to implement on 3GPP access.

Sure, I will add the corresponding text.

> Therefore, I'm not at all
> certain this sentence is really valid in wide-scale:
> --
> Finally, operators can make IPv6 as the default for always-on mobile
> connections, and use IPv4 (private) addressing only on-demand.
> --

Fair enough. I will add additional text as mentioned above.

Thanks,

-Rajeev



> 
> The parallel IPv4 and IPv6 bearer usage should anyhow be transitional phase
> until the whole network supports IPv4v6 type of bearer. Then with IPv4v6 your
> cost structure *relating to PDN connections* should be similar to just having
> IPv6? 
> 
> Best regards,
> 
> Teemu
> 
>> -----Original Message-----
>> From: ext Cameron Byrne [mailto:cb.list6@gmail.com]
>> Sent: 19. huhtikuuta 2010 02:26
>> To: Savolainen Teemu (Nokia-D/Tampere)
>> Cc: remi.despres@free.fr; rkoodli@cisco.com; fred@cisco.com;
>> v6ops@ops.ietf.org
>> Subject: Re: New Version Notification for draft-koodli-ipv6-in-mobile-
>> networks-02
>> 
>>> I think the draft, now that it is WG draft, should reflect the
>> conclusions that have been already achieved and not restart transition
>> mechanism selection discussions...
>>> 
>> 
>> When you say conclusion, you mean the conclusion from the 3GPP-IETF
>> meeting in San Francisco?  Which is dual-stack AND IPv6-only as
>> solutions?  Lets make sure the drafts currently being developed
>> reflect that to the best extent possible.  The San Francisco meeting
>> was very meaningful for me since it was the first time the larger
>> community began to grasp the notion that IPv6-only is feasible and
>> real.
>> 
>> Regarding cost per PDP, i have told the list how it is in my case, 2x
>> the cost for 3GPP dual-stack.  Look at it from a pricing contract
>> perspective or signalling load on core elements, the result is 2x the
>> PDP is 2x the cost.  I am not sure who you are talking to or what the
>> context is when you say PDP is cheap, but we all know smartphone are
>> overwhelming signalling load these days, especially PDP setup and
>> tear-down (battery saving technique).  Some service providers pay for
>> ALL of their infrastructure in the packet core based on PDP numbers,
>> just as they pay for all their circuit switched infrastructure based
>> on busy hour call attempts.  Sometimes it is a simple PDP license fee
>> mix ($5 per PDP per quarter) with capital purchases ($10 million per
>> GGSN kit).  There are many variation, but i am not aware of any SP
>> that would say 2x PDP is acceptable without 2x the revenue.  I 100%
>> assure you the cost per PDP is NOT trivial, and the increased
>> signalling from 2x the PDP attach and 2x the mobility events will be
>> substantial. Especially considering the revenue per byte is eroding,
>> service providers are very adverse to changes that drive up costs.  If
>> another network operator posts to the list and says they have a
>> different cost structure and contract where PDP is trivial (in
>> contract, in signalling, in capacity ...), great lets here from them
>> directly as another data point.
>> 
>> Ii am not interested in any new methods (DS-lite ...), they have their
>> place, but it is not appealing to the mobile scenario (handset impact)
>> and their time to market is variable.  As mentioned, i really don't
>> need any new technology to make my own path work.  IPv6-only works
>> today with NAT64/DNS64, we demonstrated to the IETF-3GPP in San
>> Francisco and our friendly user trial is growing as IPv6-only,
>> products are planned for IPv6-only.  What Rajeev adds makes things
>> clearer and defines a possible path for covering some corner cases.
>> 
>> Also, very important, Rajeev does state that the IPv6-only path exists
>> and is viable, which is something none of the other drafts on 3G/4G
>> IPv6 do to a meaningful extent.  This is a shame considering the
>> results of the San Francisco meeting.  The impression i get is that
>> most of the drafts see IPv6-only as a niche for M2M and not the
>> general Internet access case.  In saying that IPv6-only should be
>> encouraged, Rajeev is faithfully reflecting the 3GPP-IETF community's
>> official San Francisco meeting summary as well as championing the need
>> for service providers and vendors to aggressively pursue the urgent
>> need for IPv6 as a mainstream network protocol.
>> 
>> I am starting to feel like this thread is becoming an echo chamber
>> with the same people saying the same things.
>> 
>> Cameron