Redploying most of 127/8 as unicast public
Masataka Ohta
mohta at necom830.hpcl.titech.ac.jp
Sat Nov 20 01:26:33 UTC 2021
Mans Nilsson wrote:
>> With proper layering, network addresses including IP ones, certainly,
>> uniquely identify *hosts*.
>>
>> However, with proper layering, *applications* only require uniqueness
>> of IP+Port, which is enough for the worldwide IPv4 network.
>>
>> As a result, NAT won the battle against IPv6.
>>
>> IPv6 addresses are free but useless.
>
> With all due respect, you think about networks. I use and build
> networks. And my experience is that IP+port is not enough.
Certainly, local uniqueness of IP addresses to identify hosts
is required even in private networks behind NAT. But, because
of layering, that's all.
I do have extensive experiences to use and build networks
with proper layering in mind.
> We cope,
> because a lot of technical debt is amassed in corporate and ISP /
> access provider networks that won't change.
Sounds like abstract nonsense.
> We don't cope because NAT is
> good. Hardly a workday goes past without me thinking "If I could address
> this computer uniquely I'd go home earlier and with less grey hair".
The reality is that application servers only need globally unique
and stable IP+Ports.
You can address application servers with them.
> We must do better.
As IPv6 is worse than IPv4 with NAT, feel free to propose a new
network protocol.
Masataka Ohta
More information about the NANOG
mailing list