[ppml] IPv4 "Up For Grabs" proposal

John Paul Morrison jmorrison at bogomips.com
Thu Jul 12 13:56:31 EDT 2007


I really, *really* hope these systems are not connected to the global 
Internet! :-)

I think there's likely a big difference between IPv4 end of life on the 
Public Internet, and IPv4 end of life on private networks, control 
systems etc.

Right now the Internet is a bunch of private networks running IPv4 and 
internetworking with IPv4. For the Internet to evolve it will need to be 
a bunch of private networks running either IPv4 and/or v6, and 
internetworking with IPv6. (If you can interconnect with v6, v4 becomes 
redundant and will likely be thought of as a security hole).

I don't see how two protocols, side by side on the same device are 
scalable on the Public Internet for very long. We could see:

1. Stick with IPv4, forget IPv6. (Ab)use NAT, trade valuable IPv4 
addresses on the black market, the whole thing could keep going 
indefinitely but it would get more and more in the way of commerce as 
IPv4 addresses get expensive.

2. Dual stack for a transition period, but who wants to double their 
network administration workload?

Hosting farms have very complicated setups for load balancing, 
firewalls, DNS etc. I wouldn't want to keep maintaining two sets of rules.
Services providers have complicated networks, do they want to start 
messing around much if they don't have to?

If you look at 6PE for MPLS service providers, you side step having to 
make changes in the IPv4 MPLS core, and you can easily add IPv6 at the 
edge for VPNs or even the global routing table.  6to4 can do some 
similar things, leveraging IPv4 networks.
I don't really consider 6to4 or 6PE as dual stack, except at the edge.

If I operate a large hosting site and want to start offering v6 (maybe 
so developing countries can reach me natively or whatever), I'm going to 
use an appliance or firewall that automates this process, leaving 
existing systems alone for the time being.

3. "Native" IPv6 Public Internet.  I would define this as the day one 
can safely put an AAAA record in DNS as the only entry, and expect 
anyone to reach it, with the onus on the querier to deal with the 
NAT'ing to IPv4 if necessary.  At this point it becomes redundant to 
return A records. I'm sure the IPv4 Internet will still be around but I 
would assume it would mainly be carrying tunneled v6 traffic over it.



Davis, Terry L wrote:
> Jordi
>
> I agree and I started to respond to a post week with a similar response and got distracted.
>
> I can absolutely guarantee that the aviation industry expects the migration from v4 to v6 to take over 25 years.  We just expect to build airplanes that can deal with OSI, v4, and v6.  The global air traffic management system is made up of 10 of thousands of pieces controlled by approaching 1000 different organizations from small private operations to nations and v4 is already built into infrastructure pieces that are not likely to see communications upgrades for 10 to 20 years.  I routinely speak to aviation industry leaders on this and I generally place v4 end of life somewhere from 25 to 40 years out.
>
> Likewise most critical infrastructure around the globe is the same; the SCADA that runs this today is mostly all v4 as are the hospital's (including Intensive Care Units) infrastructure around the world.  This type of infrastructure is much harder to convert than just corporate IT; it takes years of planning and scores of individual governmental design approvals/certifications to change it.
>
> Take care
> Terry
>
>   
>> -----Original Message-----
>> From: JORDI PALET MARTINEZ [mailto:jordi.palet at consulintel.es]
>> Sent: Thursday, July 12, 2007 6:44 AM
>> To: ppml at arin.net
>> Subject: Re: [ppml] IPv4 "Up For Grabs" proposal
>>
>> Hi,
>>
>> I already mention this in other threads (may be not in ppml).
>>
>> IPv6 has been designed to coexist with IPv4 for an undetermined period of
>> time. It is not expected to run *only* IPv4 since day one, and not all the
>> stacks actually support this. In fact, many stacks are somehow hybrids
>> instead of two-stacks, what it means that you can't disable IPv4 (of
>> course
>> you can let IPv4 "un-configured", which is almost equivalent).
>>
>> This means that IPv4 will be here for a long time and dual-stack is the
>> main
>> transition technique. This will change with the time, at least in some
>> networks, once IPv6 traffic become predominant, among other economic
>> factors.
>>
>> You always will have, at least for many years, old IPv4 boxes that can't
>> be
>> upgrades, and the easier way to reach them is if you run dual-stack, at
>> least in the hosts in any LAN, instead of requiring translation. This
>> doesn't mean public IPv4 addresses, as in most of the situations, private
>> IPv4 behind NAT and global IPv6 will make it.
>>
>> However, the question may be different for whatever is not an end-site LAN
>> (for instance backbone, access, etc.), as there are already protocols such
>> as softwires (basically L2TP), that allow you to automatically tunnel
>> IPv4-in-IPv6 (or in the other way around today in most of the IPv4-only
>> networks), in order to be able to handle the IPv4-only applications in an
>> automatic fashion.
>>
>> This is the case for some big networks (+5.000 sites) that we have where
>> the
>> initial deployment was completely dual-stack, and then we realized that
>> because the kind of traffic was becoming predominantly IPv6, and most of
>> the
>> IPv4 traffic was basically going to Internet thru proxies, it make sense
>> to
>> turn the proxies dual-stack and carry that inside the complete network as
>> IPv4-in-IPv6 (up to the proxy), so we had been able to disable IPv4
>> everywhere (except in the LANs, for both clients and servers).
>>
>> This is the model that I certainly believe will be the one as IPv6
>> penetration becomes bigger and bigger, and then as indicated by Kevin,
>> IPv4
>> will vanish naturally ...
>>
>> I've introduced the description of this scenario also in a document that
>> I've circulated a few weeks ago
>> (http://www.ipv6tf.org/index.php?page=news/newsroom&id=3004), as I believe
>> that this will mean less trouble for possible "new" ISPs when IPv4
>> addresses
>> are gone or "almost" gone and at the same time will help existing ISPs to
>> keep growing their networks without the need for asking for more IPv4
>> addresses to the RIR.
>>
>> Regards,
>> Jordi
>>
>>
>>
>>
>>     
>>> De: Kevin Kargel <kkargel at polartel.com>
>>> Responder a: <ppml-bounces at arin.net>
>>> Fecha: Wed, 11 Jul 2007 14:07:16 -0500
>>> Para: <PPML at arin.net>
>>> Conversación: [ppml] IPv4 "Up For Grabs" proposal
>>> Asunto: Re: [ppml] IPv4 "Up For Grabs" proposal
>>>
>>> Why is there such a big push to drop IPv4?  Is there a reason that v4
>>> and v6 can't operate concurrently in perpetuity?  Won't the customers go
>>> where the content is and the content go where the money is?
>>>
>>> I would suggest that if IPv6 is a good thing (and I firmly believe that
>>> it is) then networks will naturally gravitate to IPv6.  That being the
>>> case then let IPv4 die a natural death of attrition.  There is no need
>>> to murder it outright.
>>>
>>> If in fact IPv4 continues to survive and thrive alongside IPv6 wouldn't
>>> that very fact demonstrate the need to keep it going and foster it?
>>>
>>> It sounds like a lot of people have so little faith in the value of IPv6
>>> that they for some odd reason cinsider IPv4 a threat.   If IPv6 is
>>> better than IPv4 then people will use it.  If it isn't then they will
>>> stay where they are.  I see no reason to 'force' people to switch.  They
>>> will move when it is in their best interests to do so for features and
>>> markets.
>>>
>>>
>>>
>>>
>>>
>>>       
>>>> -----Original Message-----
>>>> From: ppml-bounces at arin.net [mailto:ppml-bounces at arin.net] On
>>>> Behalf Of Ted Mittelstaedt
>>>> Sent: Monday, July 09, 2007 4:51 PM
>>>> To: bill fumerola; 'ARIN PPML'
>>>> Subject: Re: [ppml] IPv4 "Up For Grabs" proposal
>>>>
>>>>
>>>>
>>>>         
>>>>> -----Original Message-----
>>>>> From: ppml-bounces at arin.net [mailto:ppml-bounces at arin.net]On
>>>>>           
>>>> Behalf Of
>>>>         
>>>>> bill fumerola
>>>>> Sent: Monday, July 09, 2007 1:32 PM
>>>>> To: 'ARIN PPML'
>>>>> Subject: Re: [ppml] IPv4 "Up For Grabs" proposal
>>>>>
>>>>>
>>>>> On Thu, Jul 05, 2007 at 05:09:59PM -0700, Ted Mittelstaedt wrote:
>>>>>           
>>>>>>>> OK, then how exactly is this fact an argument AGAINST arin
>>>>>>>>                 
>>>>>>> simply removing
>>>>>>>               
>>>>>>>> these records out of it's whois?  Which is what I am suggesting?
>>>>>>>>                 
>>>>>>> who does that hurt? the legacy holders or the rest of the
>>>>>>>               
>>>> community
>>>>         
>>>>>>> trying to use a tool to find out who to contact when that
>>>>>>>               
>>>> netblock
>>>>         
>>>>>>> does something foolish.
>>>>>>>
>>>>>>> as a paying ARIN member, i want ARIN to keep track of as much as
>>>>>>> they're legally, financially, technically allowed to. that WHOIS
>>>>>>> service is more useful to me, the paying ARIN member, not
>>>>>>>               
>>>> the legacy holder.
>>>>         
>>>>>> For now.  What about post-IPv4 runout?
>>>>>>             
>>>>> i think you assume that ARIN's IPv4 services will change in
>>>>>           
>>>> some major
>>>>         
>>>>> way when that happens. i don't believe the memebership would
>>>>>           
>>>> want that
>>>>         
>>>>> change and the IPv6 fees at that point would cover
>>>>>           
>>>> maintanence of those
>>>>         
>>>>> 'legacy' systems.  i'd imagine ripping the IPv4 components would be
>>>>> more costly than just maintaining them after any sort of:
>>>>>           
>>>> ipv4 runout
>>>>         
>>>>> of addresses by ARIN, ipv6 eclipse of ipv4, ipv4 runout of
>>>>>           
>>>> addresses by
>>>>         
>>>>> IANA, etc.
>>>>>
>>>>> i would want to see the same level of service provided. no
>>>>>           
>>>> difference
>>>>         
>>>>> between legacy pre-ARIN holders and paid members.
>>>>>           
>>>> So then if the membership doesen't want IPv4 to be removed
>>>> from the registries, then what is going to be created is a
>>>> situation where nobody has any incentive to remove their IPv4
>>>> reachability, nor remove the ability for their customers to
>>>> reach IPv4 sites.
>>>>
>>>> In short, IPv4 will NEVER "go away"  Your proposing a future
>>>> were we add IPv6, and nobody ever gives up IPv4 resources.
>>>> So the Internet merely becomes an Internet of both IPv6 and
>>>> IPv4, not an Internet of IPv4 only or an Internet of
>>>> IPv6 only.
>>>>
>>>> I'm not debating we could or couldn't do this technically.
>>>>
>>>> However, if we do this, then don't you see that ALL IPv4
>>>> holders, not just the legacy ones, will never have any
>>>> incentive to drop IPv4.
>>>>
>>>> If all of that is OK with you, then why would an existing
>>>> paying IPv4 holder today who doesen't need numbering, want to
>>>> bother going to IPv6?  After all you just said everyone will
>>>> be maintaining their IPv4, so what need is there for an
>>>> IPv4
>>>> holder to load up IPv6?  The only incentive I see would be to
>>>> reach a network that is IPv6 ONLY, such as a network that
>>>> needs numbering post-IPv4 runout.
>>>> This puts a terrible burden on these networks because since
>>>> they are new, they cannot be reached by a lot of the
>>>> Internet, and it is not likely that they can provide enough
>>>> of an incentive to get IPv4-only holders to update to reach them.
>>>>
>>>> Ted
>>>>
>>>>
>>>> _______________________________________________
>>>> This message sent to you through the ARIN Public Policy
>>>> Mailing List (PPML at arin.net).
>>>> Manage your mailing list subscription at:
>>>> http://lists.arin.net/mailman/listinfo/ppml
>>>>
>>>>         
>>> _______________________________________________
>>> This message sent to you through the ARIN Public Policy Mailing List
>>> (PPML at arin.net).
>>> Manage your mailing list subscription at:
>>> http://lists.arin.net/mailman/listinfo/ppml
>>>       
>>
>>
>> **********************************************
>> The IPv6 Portal: http://www.ipv6tf.org
>>
>> Bye 6Bone. Hi, IPv6 !
>> http://www.ipv6day.org
>>
>> This electronic message contains information which may be privileged or
>> confidential. The information is intended to be for the use of the
>> individual(s) named above. If you are not the intended recipient be aware
>> that any disclosure, copying, distribution or use of the contents of this
>> information, including attached files, is prohibited.
>>
>>
>>
>> _______________________________________________
>> This message sent to you through the ARIN Public Policy Mailing List
>> (PPML at arin.net).
>> Manage your mailing list subscription at:
>> http://lists.arin.net/mailman/listinfo/ppml
>>     
> _______________________________________________
> This message sent to you through the ARIN Public Policy Mailing List
> (PPML at arin.net).
> Manage your mailing list subscription at:
> http://lists.arin.net/mailman/listinfo/ppml
>   

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.arin.net/pipermail/arin-ppml/attachments/20070712/12eba831/attachment.htm>


More information about the ARIN-PPML mailing list