UK among nations that have done least

[#1]

Quote from: purleigh

Don't you find it at all worrying that NOBODY understands anything you are talking about, or even what point you are trying to make ?

frustrating at times but others not so much

Quote

No !,  this has already been explained over and over !. The standard IPv6 customer allocation is a /64,  so the ISP and DNS servers only need inspect the most significant 64 bits of the (128 bit) IPv6 address to identify which customer's phone line to shove any incoming packets along. The final routing of the individual IPv6 addresses to specific machines on the LAN will be done by the customer's gateway router. Therefore the ISP only needs ONE entry in their routing table for each customer - the /64 subnet. Hence there is no need to limit the number of devices that a customer uses within their /64 (264) address range.

which has problems, like every single line would require a fully complaint BGP router (which type it is doesn't matter) for this to work, which would actually mean 2 totally separate routers in a single box if you are running dual stack and that's a lot of cost but wait there's a solution to be able to roll it out on current routers
now there was an rfc about this and how it has to be the way I described to stop things like dos attacks on end users (except of course there more likely to be multiplier attacks), even though a router in many cases on a 6 only link will be upgraded to a pure firewall it still wont be able to handle the assault, most of these problems don't currently exist because hardly anyone has ipv6 yet the problems have already been found in spades

oh and I know home routers are fairly decent for some stuff but they just don't have the processing power or the memory capability to provide all the extra services they would need to run in order to relieve the isp servers of the load for consumer addressing and still maintain full connectivity for the average number of devices on ipv6 whilst retaining the ipv4 capabilities they currently run, maybe they could do pure 6 fully but not a cat in hells chance there going to manage full 6 and full 4 without doubling the cpu + tripling the memory provisions, there fine for the tunnelled stuff because essentially its just ipv4 traffic there routing through the WAN side

[#2]

Quote from: vilefoxdemonofdoom

So a visit to Amazon would likely be followed by their spam adverts on a Google page... It would be like passing a global cookie to all web sites that I visit.

@'vilefoxdemonofdoom' - Have you tried switching OFF the Amazon "Manage Your Browsing History" ?
Doing that makes a big difference to the junk you receive, and prevents your recent Amazon browsing being copied to other people on your connection - despite them having independent Amazon accounts !
Follow -
Your Account  ->  Personalised Content  ->  Your Browsing History Settings
also check -
Your Account  ->  Personalised Content  ->  Advertising Preferences

[#3]

Quote from: jelv

Quote from: nanotm if this means I incorrectly used the label dns then so be it but i'm pretty certain its the same way that its described in all the documentation Ive been able to find Bit by bit we are getting him to admit that he actually doesn't have a clue what he is talking about!

or perhaps people are so hung up on how things work with ipv4 and nat they don't get just how different things are with ipv6 only, but then unless you actually work on a 6 only network its difficult to totally understand it, most people don't get it because there still tied to ipv4 and the way that works, because as long as the routers pushing a tunnel over the wan its not providing full ipv6 connectivity

[#4]

Quote from: nanotm

the 128bit address utilised by every piece of internet connected equipment will have to be listed in there domain name server under statefull configuration on a server by an isp in order for them to route the traffic to the consumer device from wherever it is initiated otherwise there routing equipment will blackhole the packets as destination unrecognised or the other option would be there going to be broadcasting every packet received to every consumer device (which would overload the network)

I really cannot think of any more succinct way of putting this ... that is complete and utter bollocks. There is not single statement in that which has any basic in fact whatsoever.
Really, there is nothing whatsoever in there that's correct.
DNS really, really is NOT required for IPv6 devices to be able to communicate, the situation is just the same as with IPv4. The ONLY reason DNS matters more with IPv6 is because the addresses are longer and so it'll be more of a pian typing addresses rather than using DNS.
And DNS really isn't different for IPv6 either. There's a different record type (AAAA vs A) for forward lookups, and a different zone (ipv6.arpa vs in-addr.arpa) for reverse lookups, but other than that, the way DNS works is pretty well identical between IPv4 and IPv6 - I speak from experience of having added IPv6 zones to BIND setups with no more aggro than adding the right zones and records as I'd have to do anyway.

Quote from: nanotm

which has problems, like every single line would require a fully complaint BGP router (which type it is doesn't matter) for this to work ...

See above, this is totally and utterly wrong.
The end user does NOT need BGP. The ISP does NOT need BGP anywhere near their user-side routing. The ISP may (if it suits them) run BGP inside their network, they only need to run it at the borders where they interconnect with other providers and carriers - and then they only advertise a single aggregated prefix. At the customer side, they only need to record the endpoint for each customer and the IPv6 prefix (note the singular) assigned to that customer - ie much the same as they have to do with IPv4.

I really can't imagine where on earth you get this "information" from. I can't even imagine it coming from the normal process of putting bits together, ie adding 2+2 and getting 5 - this is more like adding 2+2 and getting 57 billion !

[#5]

Quote from: nanotm

or perhaps people are so hung up on how things work with ipv4 and nat they don't get just how different things are with ipv6 only

Except that most things aren't that different with IPv6. Regarding some of the things you have hangups about, there are really hardly any differences at all. DNS isn't really any different, routing isn't that much different.

Quote

.. as long as the routers pushing a tunnel over the wan its not providing full ipv6 connectivity

Please, please stop pushing your lies. This simply is not true (like almost everything you've said in this thread). Really, it just is NOT true.
Perhaps you would explain why you believe that a process that means :

• My devices have globally unique and fully routable IPv6 addresses.


• I can pass any traffic I like between those devices and other devices anywhere in IPv6 space.


• It's in, working, and has been for some time.


• To know that it's not "native" IPv6 you'd need to do a traceroute and recognise some of the hops as belonging to HE's tunnel service.

Is in some way "not full connectivity" ?
[tt]$ ping6 www.google.com
PING6(56=40+8+8 bytes) 2001:470:1f09:126f:8440:8595:d896:6e54 --> 2a00:1450:4009:805::1011
16 bytes from 2a00:1450:4009:805::1011, icmp_seq=1 hlim=58 time=40.094 ms
16 bytes from 2a00:1450:4009:805::1011, icmp_seq=2 hlim=58 time=26.481 ms
16 bytes from 2a00:1450:4009:805::1011, icmp_seq=3 hlim=58 time=25.502 ms
16 bytes from 2a00:1450:4009:805::1011, icmp_seq=4 hlim=58 time=26.934 ms[/tt]
[tt]$ traceroute6 www.google.com
traceroute6 to www.google.com (2a00:1450:4009:805::1011) from 2001:470:1f09:126f:8440:8595:d896:6e54, 64 hops max, 12 byte packets
1  2001:470:<redacted>:d0  1274.080 ms  8163.771 ms  10775.741 ms
2  <redacted>.tunnel.tserv5.lon1.ipv6.he.net  795.812 ms  707.525 ms  927.745 ms
3  gige-g4-8.core1.lon1.he.net  4772.659 ms  26.616 ms  25.713 ms
4  2001:7f8:4::3b41:1  25.352 ms  46.480 ms  26.489 ms
5  2001:4860::1:0:3067  44.987 ms  26.454 ms  26.397 ms
6  2001:4860:0:1::4f5  29.770 ms  27.384 ms  27.091 ms
7  2a00:1450:8000:15::e  26.622 ms
    2a00:1450:8000:15::f  26.366 ms  26.170 ms[/tt]
Hop 1 is the internal interface of my router, hop 2 is the other end of the tunnel to HE. After that, there's no difference whatsoever from native IPv6.
Some of us are just doing it - not moaning about how rubbish it is because of stuff we falsely think is wrong or impossible. When I get native IPv6, nothing will change apart from a few IP addresses.

[#6]

if they only record a prefix assigned to a customer there going to be running a stateless network, that means there forwarding every packet containing the prefix regardless of the customer's actual address, this actually increases network congestion when the customer's router drops each packet and sends out a reply to let them know the endpoint is incorrect,
an analogy for you to in an attempt to portray the problem,
if I sit in my house and send a million parcels to the local village postcode with addresses that don't work my sorting office will return them to me as address not recognised this will disable my ability to send anything until I sort through it all and stop me overloading the village delivery office,
however if the sorting office doesn't record the end addresses and just forwards them on to the village delivery office that delivery centre then has to sort through each and every package and return it to the sender, while there doing this they cant process the normal mail so the people in the village get no deliveries.
if the sorting office realises that all the packages are the same they can automatically chuck them into a skip and not bother to send individual replies but instead send a bulk mail cover letter back to me informing me how much my idiocy will cost me should it occur again.
in terms of data packets the sorting office is the isp and the village delivery centre is the residential router.
in stateless auto config it has to send replies to every packet to inform the sender the address doesn't exist, in state full auto config its the isp's job to sort that out,

as to why its different when using a tunnel, the overly simplistic response is you router is pumping out ipv4 traffic to the wan so its routing rules are that of ipv4, that is also why it isn't native ipv6
as to utilising tunnel services I do that every day that doesn't mean I don't think its wholly wrong and want things made better

[#7]

Quote from: nanotm

dns on ipv6 is not the same as dns under ipv4

STOP THIS! DNS is exactly the same for IPv6 as IPv4 just using different record types as Simon has mentioned. To illustrate with an example, take my website at www.newtonnet.co.uk which sits on a server with native IPv4 and IPv6 connectivity. It has a reference in a DNS zone file as per the following:

www.newtonnet.co.uk	IN	A	85.119.83.175
www.newtonnet.co.uk	IN	AAAA	2001:ba8:1f1:f137::2

Note the only difference between IPv4 and IPv6 is that they use different record types.
However, and more importantly, you need to get our of your that DNS has anything to do with how IP packets are routed. All DNS it is a human-friendly level of asbstraction from the underlying IP addresses.
There is definitely an argument that DNS is more 'important' in IPv6 in IPv4 given that IPv6 addresses are potentially more difficult to remember/handle but there is still no absolute requirement for it.
You can see this for yourself - the following links will connect you to the website without any requirement for DNS whatsoever:

IPv4: http://85.119.83.175
IPv6: http://[2001:ba8:1f1:f137::2]

(You'll need to copy-and-paste them in to your browser - I didn't use URL tags because the forum software doesn't cope well with the use of [ ] brackets that are required when expressing IPv6 address literals.
Now, I am not suggesting for one minute that we wouldn't use DNS because as the above illustrates it's not very human friendly to use IP addresses but you need to get it out of your head that DNS is in any way involved with routing. It certainly doesn't need to get involved with connecting client devices to the Internet.

Quote

the 128bit address utilised by every piece of internet connected equipment will have to be listed in there domain name server under statefull configuration on a server by an isp in order for them to route the traffic to the consumer device from wherever it is initiated

No. This is absolutely wrong. Please stop spreading misinformation.

[#8]

Quote from: nanotm

dns on ipv6 is not the same as dns under ipv4

I think nanotm means DHCP rather than DNS, then his last post makes a bit (but not a lot) more sense...  I think he's proposing that addresses are handed out by ISP DHCPv6 servers rather than customer routers?

[#9]

Quote from: nanotm

if they only record a prefix assigned to a customer there going to be running a stateless network, that means there forwarding every packet containing the prefix regardless of the customer's actual address, this actually increases network congestion when the customer's router drops each packet and sends out a reply to let them know the endpoint is incorrect,

Under normal circumstances why would traffic be being sent to addresses that do not exist? The vast majority of traffic being sent to a residential user will be in response to a request from equipment owned by that user. Not that any of that changes the fact that routing is done by aggregation of prefixes to include the minimum required number of entries in the routing table. The only routing table that may include full addresses and not just prefixes will be the final router, i.e. the one in the user's home.

Quote from: nanotm

to why its different when using a tunnel, the overly simplistic response is you router is pumping out ipv4 traffic to the wan so its routing rules are that of ipv4, that is also why it isn't native ipv6

Yes, IPv4 is used to pass packets between the two endpoints but other than that it us entirely IPv6. The IPv4 tunnel can be disregarded when considering the flow of traffic.

[#10]

Quote from: nanotm

if they only record a prefix assigned to a customer there going to be running a stateless network, that means there forwarding every packet containing the prefix regardless of the customer's actual address

New flash for you - this is EXACTLY what happens with IPv4 as well ! ANY packet send to any* of your IPv4 addresses is simply pushed down the pipe - no regard to whether there's any device that will take that packet. If there isn't anything listening, then the packet either gets dropped, or possibly an ICMP packet is sent back (depends on network and firewall setup).
IPv6 changes nothing in this regard - packets are just sent down your pipe and the equipment at your end either acts on them or discards them. The fact that there are more addresses is barely relevant unless some tit decides to do a full ping scan. But then there's nothing to stop anyone sending a gazillion packets to your IPv4 address(es) just to fill your pipe up and disrupt your connectivity.
* Having just one IP address is in many ways a spacial case. Many of our customers have an IPv4 subnet (typically a /29 - but I manage a site with a /23 8)) and packets for all of their addresses are just squirted down the pipe.

Quote

as to why its different when using a tunnel, the overly simplistic response is you router is pumping out ipv4 traffic to the wan so its routing rules are that of ipv4, that is also why it isn't native ipv6

You still don't get it do you ? Within the tunnel, yes the IPv6 packets are encapsulated within IPv4 packets. Is that so much different to encapsulating IPv4 or IPv6 packets in ATM cells** ? or in PPP ? or with MPLS labels ? or in Frame Relay frames, or ... Pretty well most (if not all) of your traffic will have been encapsulated in such a manner during part of it's journey.
The point is that regardless of the route the encapsulated (tunnelled) packets take), outside of the tunnel (ie across the internet as a whole and within your site) they are still native IPv6 packets and no different to any other IPv6 packet.
** If you use ADSL (majority of home connections in UK) or VDSL2 (as used with FTTC) then your packets will be encapsulated in PPP and then in ATM cells (for ADSL, not sure about FTTC) - so even your IPv4 packets aren't "native IPv4" by your definition !

[#11]

Quote from: nanotm

if they only record a prefix assigned to a customer there going to be running a stateless network, that means there forwarding every packet containing the prefix regardless of the customer's actual address, this actually increases network congestion when the customer's router drops each packet and sends out a reply to let them know the endpoint is incorrect,

The prefixes (first 64 bits of the IP address) are unique to a customer, any packet starting with those 64 bits is intended to be received by something running inside the customers LAN. When the ISP receives a packet it will look at the prefix and send it to the one customer. Because every customers prefix is unique no customer will receive any packets for any other customer. If the prefix doesn't match any customer it will be bounced. None of this is any different to the way IPv4 works.
So how does that increase network traffic?

[#12]

i'd suggest you read up on multiplier attacks and mitigation /elimination of them but I'm not sure you'd be able to understand it, clearly your struggling to grasp that ipv4 is very different in operation from ipv6, never mind the concept that effectively changing ipv6 data into ipv4 and then switching it back is essentially the same as using a transformer to swap digital to analogue for transmission and then switching it back at the receiver in order to use it, that doesn't mean your connection or equipment is natively supporting the other (newer) technology it means its found a way to make it possible to use it, except of course with ipv6 and ipv4 your stripping out all the parts that make ipv6 more than just a larger address database, and seem to be forgetting that the router isn't routing ipv6 on both the wan and lan its merely acting as a pass through for the tunnel from your pc to its destination and its the pc that's putting out ipv4 packets to the router for it to send onwards, but then it couldn't very well be any thing else unless you were sharing a single ipv6 address from the router to the lan,

[#13]

Unless the end point of the tunnel is set up to be your computer, rather than your router, then your pc is sending IPv6 traffic to the router. The router is using a IPv6 routing table to determine packets should be routed to the tunnel interface. This is where IPv6 packets are encapsulated into IPv4 packets and then forwarded out the WAN interface. On the return journey the packets come in the WAN interface, are passed to the tunnel interface where the IPv4 encapsulation is removed and they are then routed using the IPv6 routing table. When ISPs offer full IPv6 connectivity the only difference will be IPv6 packets won't go through the encapsulation process.
IPv4 is very similar to IPv6 in terms of it's general operation. IPv6 was created to solve problems faced by IPv4, not re-invent the wheel. And IPv6 is not 'changed' into IPv4, it's encapsulated which is more akin to an envelope being placed inside another envelope which it can later be removed from.

[#14]

sure if your running a virtual server process for it, which I know from your post in the other thread about the cli setup is what you did,
but effectively its the same thing, the router service is only routing ipv4 data on the wan side and doesn't even know ipv6 exists
regardless of what transmission modulation it applies to that data service

[#15]

I'm not running a virtual server process. My router is running an interface which encapsulates IPv6 packets into IPv4 packets to be sent to the other end of the tunnel. My router is very much aware of IPv6, which is a necessity when it's receiving IPv6 traffic.