Withdrawl of IPv6 Technical Trial

[#1]

Quote from: purleigh

This thread is sounding more and more like this cartoon video - "Fanboy" series - IPv6 and NATs

I think both sides make their case very well.
P.S. I don't have & never will have an i-phone or other flashy mobile phone.

[#2]

Quote from: purleigh

This thread is sounding more and more like this cartoon video - "Fanboy" series - IPv6 and NATs

Hee hee - great!! 
(I shall remember that one, thanks!)

[#3]

if I said the only mobile stuff to ship with native ipv6 support was w8 devices and the latest generation of LTE devices it might indicate the scale of the problem on a uk and global scale (less than 10% of all mobile devices support ipv6) and that's also includes all those pre win8 mobile devices (which was one of the key reasons why they couldnt be upgraded to the w8 OS) and so whilst they all knew about the problem and the implementation date for ipv6 compatibility (because it was determined from a conference back in 2002) and the expected life span of those devices (up to 10 years) they all failed to act.
things are slightly different for backhaul operators (like BT/AT&T etc) because there hardware was already in situ with an expected 20+ year life span, and the standard didn't even exist.... the implementation date for ipv6 then got brought forward by 5 years and intial rollouts started on the www back in 06  (when the initial planning had suggested only after 2010  this would happen) but active compliance date was fixed as 2009 for new items (this has happened) and its only serving to compound the problems
the whole attempting to use cg nat to bypass the hardware problems in the backhaul  as an interim measure until the non compliant bits are taken out is going to end in failure, all those gamers will be unable to use new games that haven't already been added to the network profile, people in some countries wont be able to join hosts and all those co-op games that don't have a dedicated host server will no longer be useable...
one option requires huge financial outlay and the other opens them up to a huge rush of law suits... it would be cheaper and easier to just provide a dual stack solution (4+6 with bridging 4/6+6/4 permitted) at each end point and provide side by side servers providing connectivity for everything.
that would negate the need for cg-nat on a consumer basis (its called carrier grade for a reason its designed for use with backhaul operators) and it would allow retention of both address sets meaning less limitations for isp's and content site owners going forwards ...

[#4]

Quote from: nanotm

if I said the only mobile stuff to ship with native ipv6 support was w8 devices and the latest generation of LTE devices [...]

..then you'd be wrong.
Both Android and Windows Mobile have supported IPv6 since around 2009 (with Cupcake and WM6.5 respectively). I was using it personally when Windows Mobile introduced experimental support for it with WM5 back in 2008 so I'm not sure where you're getting your info from. Back then, and indeed even now in the UK, there was no IPv6 provided by the networks and so tunnels had to be used but the devices themselves provided native support. Those are the only mobiles I've had to experiment with IPv6 on so there are likely more besides that have provided native support for it for many years also.
A quick Google search suggests that many Nokia's have done so since at least 2010 when T-Mobile (in the US) started their IPv6 over GSM trial.

Quote

the whole attempting to use cg nat to bypass the hardware problems in the backhaul  as an interim measure until the non compliant bits are taken out is going to end in failure

You are missing the point of CG-NAT. It has little to do with catering for legacy hardware in the carriers network but rather the lack of (IPv4) address space for an increasing customer base that still relies on IPv4. It is not a substitute for IPv6 given that we cannot afford a Big Bang migration but is intended to work side-by-side for only as long as it is required.

Quote

its called carrier grade for a reason its designed for use with backhaul operators

This is incorrect. 'Carrier grade' is a well-defined term in the telecoms world and is in reference to the reliability and availability of any system - usually specified by five-nines availability. Yes, it is of course used by carriers but that stands to reason - it is their customers that need the address space - but the 'carrier grade' aspect of it is down to the requirement for it to work as solidly as the rest of the network given it plays such a key role in maintaining connectivity to the end user.

[#5]

I was referring to the adaptors in the devices, not the ability of the software to support 6/4 tunnels (i'm pretty sure they can all do that)
but unless all of the devices can physically connect to the ipv6 address system they wont work in a single stack networking environment......which means isp provided routers are going to need ot be either dual stack capable (which is expensive) or there going to have to continue to use ipv4 only networking

cg-nat is a pack of lies wrapped in some smoke and blown up with a whole bunch of mumbo jumbo to confuse the issue over why bt cant change there own hardware in a reasonable time frame, with bt crying they cant do it its just not monetarily feasible, then after 3 years of nonsense they decide to start trailing the convoluted ideas instead of admitting that ipv6 just isn't as easy to snoop on and they need another 5 years to figure out DPI on ipv6 over ipv6....

all a consumer needs is a dual stack router that will auto tunnel everything over the correct protocol, the wan ip pool for the uk hasn't changed and every customer takes an address allocation with them when they migrate (hence the mac) from 1 isp to another (which is why the charges are waved when you hop from one to another as well as they don't loose an address from there available pool because the new provider is saying its fine I've got that covered and its cheaper for both sides)
as for running out of ip addy's  how is cg-nat going to fix that if its not by de-allocating backhaul switches and routers from the various domain servers and changing them into port replicators (for naming convention's sake) effectively killing the whole rout advertising system at the same time as not providing ipv6 and failing abysmally at going forwards whilst hiding behind cost implications for bad planning the trials in other areas have already proved that cg-nat doesn't work and is going backwards instead of moving forwards.
I wonder if BT will publish the real world results of there trial or if they will delibertly limit the scope of reporting so as to make it appear as a healthy option on paper just the same as they made fibre optic com's appear unhealthy 20 years ago when it first became a viable long distance highspeed medium....

[#6]

Quote from: nanotm

I was referring to the adaptors in the devices, not the ability of the software to support 6/4 tunnels (i'm pretty sure they can all do that)

With respect, you really don't know what you're talking about. The 'adaptors in the devices' have supported IPv6 for at least as far back as 2008 as mentioned. It's the mobile networks that haven't.

Quote

instead of admitting that ipv6 just isn't as easy to snoop on and they need another 5 years to figure out DPI on ipv6 over ipv6....

Complete red herring, and more nonsense. DPI is a higher level concept (4-7) and is IP version agnostic.

Quote

the wan ip pool for the uk hasn't changed and every customer takes an address allocation with them when they migrate (hence the mac) from 1 isp to another (which is why the charges are waved when you hop from one to another as well as they don't loose an address from there available pool because the new provider is saying its fine I've got that covered and its cheaper for both sides)

You really don't understand how IP works do you? You can't transfer individual IP addresses between providers. There's be no aggregation and the size of the routing tables would go through the roof. The DFZ routing table currently sits at nearly 300,000 routes and this represents practically the entire 4 billion IPv4 address pool. This is already pushing the limit of what they're capable of and so further fragmentation of the address space just isn't feasible. Not that I can see any reason why you'd want to do what you propose anyway.

Quote

as for running out of ip addy's  how is cg-nat going to fix that if its not by de-allocating backhaul switches and routers from the various domain servers and changing them into port replicators (for naming convention's sake)

You're babbling now. You either genuinely don't understand the subject or I'm falling for an elaborate troll! Domain servers and port replicators?
The way CG-NAT tackles the IPv4 address depletion problem is by putting all customers behind an intermediate network. This network has a non-globally routable address space - usually 100.64.0.0/10 so as not to overlap with the typical RFC1918 ranges likely to be used by customers (10/8,172.16/12 or 192.168/16) - and all customers are NAT'ed into that space. NAT overload is used so that more than one customer can NAT onto the same CG-NAT address. This CG-NAT address is then NAT'ed, again through NAT overload, onto the (limited number of) public addresses that the ISP owns. This dual-step multiplexing enables more customers than there are public IP addresses to be supported, but breaks a bunch of protocols in the process as is well known.

[#7]

Quote from: P

I'd like to think we can do it before the UK reached 1% adoption ... ... ... It's coming. A traditional Plusnet acronym for this is RSN - Real Soon Now.

The Google graph of IPv6 adoption has exceeded 2% !
NATIVE IPv6 has reached 2.10%.

[#8]

ipv6 has a shorter time window for packet delays, dpi takes too long when intercepting packets so under ipv6 it causes a disconnect for time sensitive apps, which makes it an anti snooping protocol,
as for ip address allocation, all isp's are given a pool of them, if you cancel your broadband they loose 1 allocation of an ip (not necessarily physical) but on paper they are required to change there allocation table and the ip address then pops along into the pool of addresses available for companies to bid for.......

as for cg -nat I know what it does and broadly how it does it, all the complex technical stuff doesn't really matter that much, it broad strokes its the same as having a network within a network within a network, a process that can be run indefinitely, so long as each layer of network has a routing server, in broad concept its what they do already (customer >isp server>backhaul>data server>endpoint)  and many customers already operate double or triple network solutions on top of that as do many endpoints (which do work)
the problem with the cg-nat idea is it leaves the current network setup for the nationwide network alone and runs a per isp vpn to tunnel connections across that utilising a secondary isp server addressing xyz number of connections, this in effect removes the ability of customers to utilise pure ipv6 and forces everything into a 6>4 tunnel or 4>4 tunnel  and then bundles a thousand customers into an internal network structure through the isp server then combines a bunch of isp servers into another private network before finally accessing the WAN and being able to see and interact with anything outside, it will have a side effect of limiting peer 2 peer traffic (something many isp's desire)
it will completely not work (because unlike a corporate network routed though a data center that uses a couple of WAN IP's and allocates them through a gateway to its thousands of multi site clients (with very diminished capacity) your talking about forcing millions of people into a system that wont provide the connectivity throughput to cater for them all.
the trial didn't address this issue at all but it did prove that its possible and with some tweaking it can be cobbled together to work (and BT knows that without some serious quantum routing capable of 30+ terra byte speeds the average customer if everyone is under cg-nat will see themselves achieve near 512 connections when they just sold everyone megabit fibre....

PlusNet however could actually very easily change there isp servers to utilise ipv6 to address there consumers (ok the in home routers would need to be messed about with but i'm sure they can push firmware for the majority) instantly freeing up thousands of ipv4 addresses, meaning they would be at liberty to sell there address pool on.  they would still need to maintain about a thousand and everyone who connected to a 4 only website would have to be tunnelled through one of them (but it is mostly web browsing)

the use of cg nat to effectively create new batches of non addressable ipv4 doesn't work, when they needed to come up with was open networking, to channel each batch of a thousand consumers through a tertiary hop of open nat routeing, as things stand right now cg-nat wont work on a large scale, because the data pipes and the speed of routing on a large scale just arnt big enough to cope. hooking multiple gate ways together in a daisy chain though was never going to work, each layer needs a dns server, each layer adds 30+ms, and there talking about adding 10 layers "to make it stable" at a national level sounds far more like there gearing up for isolation from the world ala Burma than anything else, gaming wont work outside of the uk, peer 2 peer wont work outside of the uk ....totally barking mad idea  

[#9]

Quote from: purleigh

In the coming few days, the Google graph of IPv6 adoption will exceed 2%, and has moved on considerably since last year. Other countries leading the way include - USA=4.12%, France=5.11%, Germany=4.17% Looking at the shape of that exponential graph, and the rate of increase in NATIVE IPv6 adoption, IPv6 MUST happen soon

the high adoption rates are due to things like Skype and other Microsoft stuff running on ipv6 hosts and so if you want them to work fast and flawless you have to be on 6 also ....(even if 90% o the middle is still running on 6/4 hops)

[#10]

Quote from: nanotm

as for ip address allocation, all isp's are given a pool of them, if you cancel your broadband they loose 1 allocation of an ip (not necessarily physical) but on paper they are required to change there allocation table and the ip address then pops along into the pool of addresses available for companies to bid for.......

Complete and utter tosh!
IPs are allocated in blocks. To allocate individual IPs to ISPs would make the routing tables a nightmare!

[#11]

Quote from: nanotm

ipv6 has a shorter time window for packet delays, dpi takes too long when intercepting packets so under ipv6 it causes a disconnect for time sensitive apps, which makes it an anti snooping protocol,

Nonsense. Shorter time window for packet delays? What are you on about?

Quote

as for ip address allocation, all isp's are given a pool of them, if you cancel your broadband they loose 1 allocation of an ip (not necessarily physical)

Nonsense. You really don't know how IP works do you?

Quote

but on paper they are required to change there allocation table and the ip address then pops along into the pool of addresses available for companies to bid for.......

I give up. You don't have a clue what you are talking about and I don't have the patience to correct your misunderstandings any further.

[#12]

Quote from: MJN

I give up. You don't have a clue what you are talking about and I don't have the patience to correct your misunderstandings any further.

   
@'MJN',  you lasted longer than me !.  
I had given up after the reply I got about the Billion 7800 IPv6 configuration, as it was clearly not worth asking for further clarification if the reply was going to be yet again irrelevant or shrouded in nonsensical technical terminology.

Having read 'nanotm's responses in this and other threads, they all seem to be phrased (in his OWN words) as -

Quote from: nanotm

wrapped in some smoke and blown up with a whole bunch of mumbo jumbo to confuse the issue

[#13]

[Moderator's note by Jim (Oldjim)  I would remind you all of this Forum Rule and this Forum Rule either of which may apply.
So please stop it ]

[#14]

Fear not Jim, I have pulled out what little hair I have left on this one and so shall avoid breaching the rules entirely by not replying to my learned friend.
I can assure all concerned that any crossing of the line has been down to sheer frustration and is nothing personal.

[#15]

There is still the problem that ipv6 breaks a number of protocols.