11
Leave a Reply

avatar
11 Comment threads
0 Thread replies
4 Followers
 
Most reacted comment
Hottest comment thread
8 Comment authors
Joe TouchJean-DanielnobitakunSanderwilly Recent comment authors
  Subscribe  
newest oldest most voted
Notify of
FransM
Guest
FransM

I’m lost a bit on what this solves. Legacy IPv4 systems do not generate this IPv10 header
And IPv4 addresses can be used in IPv6 addresses (by using a 0:0:0:0:0:FFFF prefix)
and there is 6to4.

If you need to change anyway then why invent a new variant. Why not go to IPv6?

TC
Guest
TC

So it’s like NAT but without the fixed port forwarding stuff?

willy
Guest
willy

I’ve read the entire spec and it’s IMHO ridiculous as it attacks the wrong problems by creating new ones. The first reason of low deployment of IPv6 is lack of incentive. Your home appliances don’t use IPv6 because it’s pointless since your ISP will not support it. And your ISP doesn’t care about investing money to maintain a full blown IPv6 infrastructure since people don’t have any IPv6 appliances at home. This is what has fossilized IPv4. And yes I do have IPv6 at home and it has even helped me quite a few times.

Here he’s proposing to wait for 20 more years for his new IP protocol to be specified, implemented, debugged, attacked, fixed and deployed everywhere. Also he reproduces some design mistakes of IPv6 : flow label is useless and wastes 20 bits (cf RFC7098). NextHeader is a gross design error that makes IPv6 headers unparsable by those who don’t know the exact size of all possible 256 header types since there’s no length information. TCP was smarter at this as it prefixes them with a length. Addresses contain MAC addresses (which remain a pain in IPv6 on the server side), and even has provisions for ASN (easier routing but will not be known in advance).

So in the end, he will need to have IPv10 enabled in a contiguous chain between the client and the server so that the last hop may decide to transform the protocol to IPv4/IPv6 (without knowing it’s the last one). Clients will not use it not knowing if the frames will get dropped by lack of infrastructure support. Server side will not enable it by lack of hardware support forcing them to disable all hardware acceleration. Infrastructure will not enable it by lack of users.

That’s exactly how you add a 3rd implementation of something when trying to converge 2 existing ones.

I think it would make more sense to encourage to natively *route* ipv4-in-ipv6 to real IPv4 addresses so that IPv6-aware clients have higher chances of reaching IPv4 hosts and that IPv6 on the client side is no more a dead end.

Sander
Guest
Sander

What were they thinking?! “After 20 years IPv6 is still deployed too slowly, so let’s come up with yet another protocol to de-focus from IPv6”

As @willy says “The first reason of low deployment of IPv6 is lack of incentive.”. Indeed.
Maybe that incentive will change if/when the price for IPv4 addresses rises (currently about 10 Euro per IP address). If not … so be it … let the market decide.

And, yes, I have IPv6 at home and on my VPSes.

nobitakun
Guest
nobitakun

Sander :
What were they thinking?! “After 20 years IPv6 is still deployed too slowly, so let’s come up with yet another protocol to de-focus from IPv6”
As @willy says “The first reason of low deployment of IPv6 is lack of incentive.”. Indeed.
Maybe that incentive will change if/when the price for IPv4 addresses rises (currently about 10 Euro per IP address). If not … so be it … let the market decide.
And, yes, I have IPv6 at home and on my VPSes.

I don’t totally agree, this does not pretend to substitute anything, it’s just a container with a translator between IPv4 IPv6 masked as a new protocol, but it is not. The inexistent headers in the IPv4 mean nothing since IPv10 will create its own header, and the translation will be done in a different way than NAT64, hence more efficient. In other words, this is just a faster workaround than the current solution, so it’s welcomed.

Anyways, I hope by the end of this decade half of the devices / providers in the world start to take IPv6 as a standard. Maybe I’m too optimistic lol…

Sander
Guest
Sander

Start reading (and laughing) here: https://www.ietf.org/mail-archive/web/ietf/current/msg100261.html … the author, and the reactions.

Jean-Daniel
Guest
Jean-Daniel

@cnxsoft

And how is the IPv4 system supposed to put the IPv6 destination address in the packet header ?

Joe Touch
Guest
Joe Touch

@cnxsoft is correct. This is a proposal by an individual. The IETF allows anyone to propose anything. There have been past proposals for new math that “proved” that 2+2 != 4, for example.

Unfortunately, this proposal also self-assigns a version number (10) – even though IP version numbers are assigned much later in the standards process, only when there is sufficient community support within the IETF for the development of a new standard. The proposal is now being called “IPmix”, FWIW, to avoid that problem.

If you have questions for the author of this document, you should contact them directly.

Back to top