websockets
need anyone say more
The UK's first trial of mobile IPv6 is suffering technical delays as the operator struggles to make it work, though that's still impressive when the competition can't work out why they should even bother. Mobile phones themselves have supported IPv6 for the last five years or so, but network operators have almost nothing to …
Websockets are OK, but there are still holes.; eg, the spec isn't finished yet, there are security considerations, wire level binary isn't supported (you end up with a UTF8 text encoding instead).
Lastly, while they allow duplex connections, the actual connection can only be established in a single direction, So you cannot have a 'server' connect to a 'client', meaning that you end up with lots of very long lived connections. This is fine for many situations, and an improvement on simple polling, but its still a bit of a bodge job in my opinion.
Better than the long polling or forever frame hacks you have to do right now to get decent latency in a browser app get with HTTP, but still not as good as proper sockets would be.
...it must be a good idea bypassing all those firewalls that block me when I try to e-mail a confidential attachment to my web mail account. It is work related.
/IRONY
Look at the APP DB's for some security vendors to see how wide the "Port 80" tunnel is already - and why BOFH's cannot afford to ignore this attack vector for their networks.
If you have connectivity without security, you will rapidly end up with no connectivity or even trousers as the criminal hackers pwn your PC, install a keylogger, send SPAM, empty your bank account and end your marriage.
Mine's the one with the BackTrack CD in the pocket...
As the last few octets (yes I know this isnt the defact name yet, but you know what I mean) of the IPv6 address is from the MAC address of the device, surely it will make it easier for operators to know what device is where.
Also it'll likely help them link mobile contracts with mobiles, tablet contracts with tablets
Not necessarily great for the users (based on the way we get screwed either way) but from a billing perspective it must better for the ops
The IPv6 address does **NOT** have to include the MAC address and never has. It's just one technique for easily self-assigning an address that doesn't clash with anything else on the network. Pretty well any device should support privacy mode where the address is chosen at random, and of course there are things like DHCP for IPv6.
In a mobile network, there's no reason for the operator not to allocate the IPv6 address a device gets - which could be permanent, or dynamic (changes each time the phone establishes connectivity).
Until now, the limit of my (admittedly half-hearted) wondering about whether my mobile is IPv6-capable, was concentrated on its WiFi connectivity - in other words, when my home network eventually goes over to IPv6, whether I'd be able to access it "natively" via my mobe, or have to rely on the router to provide v4-v6 bridging.
I'm still not quite sure about that (but then again, seeing as our Home Hub 3 doesn't seem to be IPv6-enabled, it's not as if I'm in a hurry), but I now realise I hadn't thought about whether/when/how the mobile networks themselves would handle the move to IPv6.
If this article is correct, it sounds as if THEY'VE been trying not to think about it...
Most of the home routers out there don't support IPv6 (indeed, many of them don't even support important IPv4 features like multicast) so I confidently predict the vestiges of the 20th Century Internet will be with us for some time to come and be the consumer norm for the foreseeable future.
The way to address the scaling of servers, assuming each phone has a fixed IPV6 address, is for the phone to register once with a server and exchange keys. Then if the server has traffic it can fire off a ping with its ID and something encrypted with the key to a defined common port on the phone. The phone then looks at the packet, checks its validity and notifies whichever application(s) on the phone have registered to receive notifications from the server, at which point they can connect and you've achieved the same as polling. The overhead of authentication each time is probably less than the overhead of all those polling packets flying back and forth. There is nothing to stop multiple registrations to multiple servers which can all send their packets to the common port. Telcos can continue blocking inbound stuff to mobiles provided they let through packets (in a standard format) to that one port.
There we are, unless someone has patented it already, it's in the public domain so go use it.
"Today's mobile operators don't provide real (internet) IP addresses to their users - customers of a single network operator will share two or three internet IP addresses between millions of users."
Some do, but more importantly, you CANNOT have millions of users using just two or three real IP's concurrently. TCP/UDP port range only goes up to 65536, so the most number of concurrent sessions is less than that. A NAT session is also not instantly torn down either.
With the government talking about mandating a access controls on porn and ensuring all porn goes on a designated domain, why not kill two birds with one stone? Insist all porn sites have to go on IP version 6 addresses. Then the switch with everyone will ensuring their kit is compatible will happen quicker than most will care to admit :)
The MAC address does not have to be used to derive the host portion of the IPv6 address. This is for autoconfiguration on LAN's only.
Also, IPv6 has privacy options that can automagically change the EU IP address to spike the logs of webservers / ISP's.
Most importantly, the IPv4 / IPv6 addresses issued to handsets are dished out by the GGSN, not by a DHCP server or autoconfiguration. The telco is aware of the IMSI and MSISDN *before* the IPv4 / IPv6 address is allocated. Check the specs for GTPv1 release 9 on the 3gpp web site.
*THEN* RTFM on MIPv6 and you realise the scale of the problem...
This post has been deleted by its author
I went to a public meetup with the CEO of one of the newer Canadian carriers, Mobilicity, the other day, and as it was World IPv6 Day, I asked him what Mobilicity's IPv6 plans were, expecting either a blank stare, a 'huh?', or some bluster. He came out with a pretty detailed yet clearly off the cuff reply, which was impressive. Sadly the answer boiled down to 'we're working on an implementation plan but we have no firm roll-out schedule yet', but hey, it was nice that the guy clearly knew what the hell he was talking about...