[180367] in North American Network Operators' Group
Re: AWS Elastic IP architecture
daemon@ATHENA.MIT.EDU (Matthew Kaufman)
Mon Jun 1 15:34:08 2015
X-Original-To: nanog@nanog.org
Date: Mon, 01 Jun 2015 12:34:04 -0700
From: Matthew Kaufman <matthew@matthew.at>
To: Christopher Morrow <morrowc.lists@gmail.com>
In-Reply-To: <CAL9jLab3XTNCYwn53v7o6__P9EO9Rv42eVAHNkf1rWNcX93Amw@mail.gmail.com>
Cc: nanog list <nanog@nanog.org>
Errors-To: nanog-bounces@nanog.org
On 6/1/2015 12:12 PM, Christopher Morrow wrote:
> On Mon, Jun 1, 2015 at 1:49 PM, Matthew Kaufman <matthew@matthew.at> wrote:
>> 1) An all-IPv6 network inside, so the hosts can all talk to each other over
>> IPv6 without using (potentially overlapping copies of) RFC1918 space...
>
> this point keeps coming up... I don't see that 'overlapping ipv4'
> matters at all here. it is presented to the customer (vm oeprator) as
> 'a flat-ish lan' where you poke from machine to machine via names.
>
> (so it seems like a rathole/FUD-problem we can just stop talking about now)
>
> -chris
I have deployed services in clouds where the overlapping RFC1918 space
did present challenges to the software stack that was trying to exchange
node reachability as IP/port. So yes, there were and still are cases
where existing software that is not aware of potential overlapped
assignments can break.
Matthew Kaufman