[LINK] IPv6 doomed

[Frank O'Connor]

At 12:41 PM +0800 24/8/08, Adrian Chadd wrote:
>On Sun, Aug 24, 2008, Frank O'Connor wrote:
>>  Mmmm,
>>
>>  Good pick-up.
>>
>>  IPv6 has a number of other advantages over IPv4, other than address space.
>
>Most of which don't really matter today.


And a few which really do.

>
>The useful ones (eg mostly-pervasive IPSEC) were ported to IPv4.
>
>The rest of the world widely deployed DHCP; autoconf sounds great in theory
>but seems to be missing 10 + years of commercial DHCP development/features.

Horses for courses. Just had an experience installing a new router 
that I could have really used IPv6 auto-config for. (Basically the 
router manufacturer provided a 'user friendly' front end app for 
doing same ... just click and it does the work. Yeah ... right! Three 
hours later after I'd undone all the damn garbage it had wreaked in 
the settings and manually installed the damn thing - which I should 
have done manually in the first place.)

>
>Noone has figured out how to "do" v6 inter-domain routing any way other
>than the "v4 plus more bits" method.

Possibly because they didn't evisage the need for inter-domain 
situations. (Probably because the address space, IPSEC and other IPv6 
features meant they'd be technically unnecessary ... simple geek 
idealism without accounting for the paranoia, empire building, 
politics and sheer bloody mindedness of our wonderful species)

>
>I reckon networks will be doing traffic engineering the same way 
>they do it now -
>including the "break up our IP space into lots of smaller chunks and 
>treat each
>seperately" trick which deaggregates the space just like it is in IPv4.

Ahhhh ... domaining ... an idea whose time has come? (Again, and 
again, and again and again ... and which invariably leaves everyone 
in the hands, and under the whims, of system admins and hard coded 
network locations ... that are less than useful in todays fast moving 
and currently geo-differentiated - in a network sense - world.)

>
>Also, some friends of mine who work for ISPs in places like Finland note
>that their commodity internet deployments involve "bridge" CPE equipment
>which just pass the packets through like an ethernet switch. They assign
>IPs and run DHCP on the ISP side. His notion of IPv6 is "assign users
>individual IPs and potentially charge for the number of computers per house";
>thats apparently their business model now and it works very well.

And because we have NAT we are technically more competent, ethically 
more valid, and way more efficient and righteous? (The guys at the 
stable must have been working on similar arguments when the 
automobile appeared.)

Seems to me that charging is what the telecommunications world is 
about ... and what naffs me is that what they charge for some things 
that would be a breeze in an IPv6 environment (high speed broadband 
laptop, PDA and phone access when overseas or travelling) is often 
bloody astronomical in an IPv4 context.

>
>So effectively it -is- about the extra address space. Maybe some markets have
>picked up the "v6 features" like mobile; there's fun anycast possibilities in
>v6 which you're slightly more restricted to w/ v4, but noone has recently put
>forward any kind of benefits re: ipv6.

Right on, Adrian ... Little numbers like its way more capable (and 
way more efficient and effective in terms of bandwidth usage) data 
casting capabilities, added bits for packet descriptors, QOS 
capabilities and the like seem to me to be rather useful .... but 
maybe that's just me. Sure we can re-engineer IPv4 (usually at the 
rather insecure transport and application layers) to patch up it's 
annoying deficiencies ... as we did with NAT and DHCP and the like 
... or we can start with a core that was specifically built to 
overcome those deficiencies that necessitated all the damn recoding 
in the first place, and use the facilities inherent in it.

Of course the other really cool thing about all that additional 
address space is that little numbers like VOIP, mobile conferencing 
and the like would/could easily become pervasive and independent of 
geolocation, network infrastructure and other current lock-ins of our 
(still!) hardwired basically twisted pair telecom providers ... which 
of course is another reason our current telecom providers (and the 
ITU, and other bodies) haven't exactly supported IPv6 and its 
capabilities.

Bottom line: IPv6 hasn't been adopted for any number of reasons . I 
mean, can you imagine anything so silly and containing so many 
overheads and inefficiencies, as to tunnel one network protocol into 
another obsolete one to make it work, it can't be handled natively on 
90% of the planet's current networking hardware, it has very few cool 
admin utilities like plethora available in the older more mature 
networking standard of IPv4 and its lack of history as a mature 
networking standard count against it. Factor in that there's a lot of 
commercial and government bodies with a heap of vested interest in 
IPv4, security regimes in place that doesn't want its ability to tap 
into the data-flows of individuals messed with, and whole industries 
with vested interests in the status quo ... and things won't proceed 
real fast.

But hey, IPv4 was in the same situation 30 years ago ... and 10-20 
years ago the telcos were caught on the hop by packet switching ... 
but IPv4 is now way long in the tooth and showing signs of breaking.

I think we will (eventually!) move to IPv6 ... but am not holding my breath.

					Regards,