Re: [105attendees] Why do we need to go for 128 bits address space?

On 28-Jul-19 09:34, John Day wrote:
> All of this talk of backward compatibility is pretty rich, considering the strong backward compatibility between IPv4 and IPv6.

To be clear, backward compatibility in the strict sense was *never* a goal for IPng, because of the lack of address extensibility in RFC791. The goal was co-existence from the start, which led inevitably to the dual-stack model and to one form or other of NAT64. 

Deploying IPv6 to the point where by some measurements it exceeds 25% of traffic has taken 25 years. So I don't think anybody will be investing in another such effort any time soon.

    Brian

> 
> 
> John
> 
>> On Jul 27, 2019, at 17:26, Simon Hobson <linux@thehobsons.co.uk> wrote:
>>
>> shyam bandyopadhyay <shyamb66@gmail.com> wrote:
>>
>>> I am thankful to you because you are the only one who came up
>>> with all the queries earlier. I am attaching a text version of the mail that I
>>> had sent which contains answers to your questions (in case 
>>> you failed to open the attachment that I had already sent).
>>
>> All you have come up with in that text is "most of that can be done in 64 bits" - but you haven't stated any case for WHY only 64 bits would be significantly better. There are few (and getting fewer) systems/situations where the extra bits needed for storage and transmission are going to be a significant issue - and many systems where it could be an issue (deep space probes ?) tend to use custom protocols anyway. What's more, the case for the change would have to be very VERY compelling given that we've now had a couple of decades of design, development, and implementations of IPv6 protocols using 128 bit addresses.
>> To change it now would be VERY problematic - a 64bit version would be incompatible with what's already been implemented by hundred or thousands of vendors across thousands or even millions of networks. So there would be all the problems that are being discussed & solved in migrating from IPv4 to IPv6 without a "global big bang day" - but (AFAICS) really no compelling case for making the change at all. You'd effectively be saying "lets forget about that 128bit IPv6 everyone has been working on, and do an IPv8 instead with only 64 bit addresses" !
>>
>> I think you need to realise that this train left the station a couple of decades ago (that's when address length was discussed and decided upon), and its' not coming back,
>>
>> --------------------------------------------------------------------
>> IETF IPv6 working group mailing list
>> ipv6@ietf.org
>> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
>> --------------------------------------------------------------------
> 
> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> ipv6@ietf.org
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------
> 