Re: [RAM] [Fwd: I-D ACTION:draft-carpenter-idloc-map-cons-01.txt]

[Iljitsch van Beijnum <iljitsch at muada.com>]

On 5-jun-2007, at 9:28, Brian E Carpenter wrote:

> 	Title		: General Identifier-Locator Mapping Considerations
> 	Author(s)	: B. Carpenter
> 	Filename	: draft-carpenter-idloc-map-cons-01.txt

The trouble I keep having with all of this is as follows.

We have identifiers in the form of fully qualified domain names.  
There doesn't seem to be a wide-scale dissatisfaction there.

Then we have locators in the form of the values found in the  
destination address field of IP packets. Although not everyone is  
completely happy about how those work, we've managed to build  
infrastructures that work with these locators very effictively.

The problem is in the relationship between the FQDNs and the  
addresses in IP packets. For reasons we all know (and probably a few  
that we haven't stumbled on yet) it has proven very hard to remap an  
FQDN to a new IP address if, when and as fast as required.

So fundamentally, the problem that we have won't be solved by another  
layer of identity somewhere in the middle. The only thing that that  
does is hide what we need to do from layers that may be hard to  
change. This means that such a layer is an engineering compromise  
rather than an architectural first class citizen.

_______________________________________________
RAM mailing list
RAM at iab.org
https://www1.ietf.org/mailman/listinfo/ram