Re: Definition of Registry

[Marcel Schneider <schneider@switch.ch>]

On Friday, 5 Jan 2001, Patrik =?iso-8859-1?Q?F=E4ltstr=F6m?= writes:
> At 11.38 +0100 01-01-05, Marcel Schneider wrote:
>>On Friday, 5 Jan 2001, "Peter Mott" writes:
>>
>>I completely agree with Peter's analysis blow. But the
>>important fact for this group is: there are two models
>>for registries/registrars/agents. The 'lightweight'
>>registry model (just DB and connectivity) is mostly
>>used in gTLD's and will continue to be (one of the
>>reasons is that it has not the registry as a bottelneck).
>>
>>The 'policy-setting' registry is more common in ccTLD's and
>>will continue to be.

> Isn't it the case that we actually have two orthogonal things we talk about?

Think so too.

>                | no-policy   |    policy   |
> --------------+-------------+-------------+
> thin registry |             |             |
> --------------+-------------+-------------+
> fat registry  |             |             |
> --------------+-------------+-------------+

> I.e. the question on thin or fat is just about where information of 
> domain name holder is (at registry or registrar) while policy can 
> still exist or not, regardless of this fact.

Maybe we could put it in different wordings:

A 'smart' registry checks policy.
A 'dumb' registry does not check policy.

A 'fat' registry registry contains a most complete set of
   domain name information.
A 'thin' registry contains only a minimal set of domain
   name information, usually the domain name - registrar
   tuple and creation/modification data.

This means we do not have to care about smart or dumb 
registries because this is policy issues only.

We also do not have to care what is published by
registrars/agents/registries because this is privacy
stuff.

The rrp should be designed that it could carry a full set
of registry data as required by a 'fat' registry: all
contacts, name servers, billing info, ceation and mod
dates (maybe more, we will have to determine this) and
all of the control information. The interaction of 
registrars with 'thin' registries will just be a 
subset of the above (hopefully ;-)).


> Yes, most domains _with_ policy are also fat registries, but that is 
> a statistical fact, and not something built into the definitions.

Right.

> Also, you can have a registry for non-TLD's aswell, aswell as for 
> "other things than domains" as Karl said. If the consensus on this 
> list were to do something for domains, and not too generic, even 
> though it is good if it can be used for other things, I still think 
> we should talk about a registry being the responsible registry for a 
> domain, and not just TLD.

Also my opinion. If Karl could supply reasons for a different
approach (a more generic) we'd be glad to reconsider.


Marcel