[ppml] Policy Proposal 2005-2: Directory Services Overhaul

Edward Lewis Ed.Lewis at neustar.biz
Mon Apr 18 14:38:01 EDT 2005

At 15:04 +0100 4/18/05, Michael.Dillon at radianz.com wrote:

>ARIN should never publish any contact
>information for an organization which does not explicitly
>agree with ARIN that its contact information should be published.

There's something intriguing in that statement.

Having been exposed to the way gTLDs are run under ICANN, I have come 
to see that the relationship between registries and registrars 
conforming to the shared registry model has some relevance to the 
relationship between ARIN and ISPs, or RIRs and NIRs/LIRs/ISPs in 
other regions.

One of the aspects of the relationship is where information is 
collected, stored and published.  In the shared registry model, all 
information from registrants is collected by registrars, there is no 
direct (business) relationship between the registry and registrant. 
This is akin to "provider assigned" address space.

It's important to keep in mind that not all name registries operate 
according to ICANN's shared registry model.  Some registries still go 
directly to the registrant, some use a mix.  In some cases, there's 
the concept of a "registrar of last resort" offered by the registry. 
The latter is akin to the legacy or "provider independent" space.

(I may be glossing over crucial details...)

Coming back to the comment above, one of the design points in shared 
registries is whether the contact information, etc., is maintained at 
the registrars and only the DNS-relevant (for lack of a better term) 
is propagated to the registry.    This is called a "thin registry." 
A "thick registry" means that all information, contact, DNS, etc., is 
propagated to the registry.

There are debates as to which approach is better.  I'll pick on one 
advantage of thick registry though - that all of the information 
collected is in one place, which is a benefit if one of the 
registrar/ISP goes out of business.  In the name arena, this could be 
significant.  Maybe this isn't as important in the number arena - if 
you're ISP goes belly up, does it matter that you retain your 
provider assigned address space?  (Folks might say yes.)

What this might lead to is the idea that ARIN collects all 
information from the ISPs, as it does now.  This relates only to 
collection though - I haven't addresses publication, aka disclosure, 
which is the hot button.

It could be that ARIN collects everything, but only in an "escrow" 
function.  Responsibility for disclosure of "down streamer" 
information may lie with the ISP - meaning now that the ISP has to do 
this (or somehow relies on ARIN for this).  WhoIs hasn't been good at 
referrals, this is where a tool like IRIS comes in.  (I won't dignify 
RWhoIs here. ;))

Besides being able to support a thick or thin, or a combo thick/thin 
model registry, IRIS is also able to support richer disclosure 
policies.  I think that is also an important point for ARIN.

This is just one aspect of the directory service work to be done. 
IRIS needs some work - the IETF hasn't finalized the document 
defining the schema for address registries.  The basic protocol doc 
is done, as well as the domain name schema.
Edward Lewis                                                +1-571-434-5468

If you knew what I was thinking, you'd understand what I was saying.

More information about the ARIN-PPML mailing list