[ih] vm vs. memory

Brian E Carpenter brian.e.carpenter at gmail.com
Tue Oct 24 12:22:36 PDT 2017


On 25/10/2017 07:09, Toerless Eckert wrote:
> Yes, these are the classical arguments. 
> 
> IMHO, arguments 1. and 2. have mostly failed, especially
> in large enterprises. They only provided some hard shell to the
> outside, but mayority of attacks can easily come from the inside.
> And protection to the outside has evolved long ago from trying
> to (unnecessarily) hiding your addressing structure over
> to app-level - keep the good bits in, and the bad bits out.
> 
> Argument 3 (i think you mean access providers) is more interesting.
> 
> I would love to hear from folks more involved in current deployments
> what the BCP is for organizations using provider dependent
> addresses to be able to quickly switch providers - without NAT.

This is hardly history (except for how we got into this mess**)
but the answer is probably RFC7157 plus RFC8028, with RFC4192
and RFC7010 in the background.

**https://www.cs.auckland.ac.nz/~brian/CCR-201404-IPaddrHarmful.pdf

   Brian

> I guess you would effectively build all org internal addressing & naming
> on ULA, and use the provider addresses only for internal<->external
> communications, but if you have an actual L3 network in the org, then
> there is probably still a lot of renumbering necessary for which
> there are no well defined network wide autoamted solutions. Although
> i think there will be a new WG, forgot name to start tackling this.
> 
> If it was me, would have just evolved and improved on rfc1928.
> 
> Cheers
>     Toerless
> 
> On Tue, Oct 24, 2017 at 07:35:12PM +0200, Paul Vixie wrote:
>>> On Tue, Oct 24, 2017 at 02:12:06PM +0200, Paul Vixie wrote:
>>>>
>>>> ...
>>>>
>>>> LISP may be an example. NAT certainly is.
>>
>> Toerless Eckert wrote:
>>> Hmm... what are the redeeming qualities of NAT ?
>>
>> every other attempt to add rapid renumbering and transparent
>> multihoming has been rejected. NAT, by not trying to do those things
>> and by not saying it would do those things, snuck under the
>> defenses.
>>
>> no multi-national enterprise should give real external addresses to
>> all of its internal endpoints, for at least three reasons:
>>
>> 1. the internal structure should not be visible or guessable.
>>
>> 2. reachability should be prevented by more than just firewalls.
>>
>> 3. you can add and drop transit providers as often as you want.
>>
>> NAT did that. nothing else could have or did.
>>
>> -- 
>> P Vixie
> 



More information about the Internet-history mailing list