Re: [irtf-discuss] Why do we need to go with 128 bits address space ?

[Phillip Hallam-Baker <[e-mail]>]

On Sat, Aug 17, 2019 at 8:29 PM Michael <[e-mail]> wrote:

> That was not very helpful useful informative responsive etc.  how about
> identifying what you think is incorrect and why?
>

I think folk are more interested in being seen to be correct rather than
making an argument. I also had someone 'correct' me when I pointed out that
the lower 64 bits were to map to a MAC address which was in the process of
being extended to 64 bits by being told that lower 64 bits were to map to a
MAC address which had been extended to 64 bits

In practice we have 2^32 routable networks in the Internet because that is
what BGP supports. We do not route on IP addresses, that hasn't happened
since the earliest days of the net.

Whether packets have 256 bits of routing information (128 bit source +
destination) or 128 is really not a major efficiency issue. Changing it
would require an absurd amount of effort for no real gain.

What would make a major efficiency improvement is to use super jumbo
frames. IP packets can be up to 64KB. Now that 1Gb/s networking is
ubiquitous and 10Gb/s is on its way and memory is cheap, we should just
bite the bullet and expand the MTU to minimize the amount of time routers
have to spend doing switching.