Re: EPP reliance on registrar sponsorship model

[Daniel Manley <dmanley@tucows.com>]

Again, this is probably too specific to the TLD registries and 
registrars, but allowing any registrar to update registry objects as 
long as the "auth info" is known would be a headache.  Would renewals be 
allowed?  And we'd have to go on word of honour that the non-sponsoring 
registrars don't keep a copy of the auth info for objects they don't own?

Registrars would almost have to build two systems:  one for their loyal 
customers and another to deal with the cusomters of other fly-by-night 
registrars that collect registrations dirt cheap but don't provide 
adequate maintainance functionality.  This would be a big burden on 
registrars and could be confusing to registrants.

Dan

Hollenbeck, Scott wrote:

>>-----Original Message-----
>>From: Jarle Greipsland [mailto:jarle@uninett.no]
>>Sent: Wednesday, August 08, 2001 7:16 AM
>>To: ietf-provreg@cafax.se
>>Subject: EPP reliance on registrar sponsorship model
>>
>>
>>I think we would be better served by having the registrar sponsorship
>>concept replaced by a more general authorization mechanism.
>>Registries that wish to work within the registrar sponsorship model
>>can then do so by (explicitly or implicitly) authorizing the "current
>>registrar" to make any changes to a given object, including the right
>>to reject or accept a transfer.  Registries that wish to operate
>>within another model, e.g. the registry will register digital
>>certificates for individuals or organizations, and attach attributes
>>to objects authorizing certificate holders to perform a given set of
>>operations on the object (provided the request is properly signed by
>>the holder) no matter what registrar forwards the request to the
>>registry, can then do so.
>>
>
>The EPP drafts were first written without object-specific authorization
>information.  It was added in an attempt to address one person's comments.
>It was subsequently removed as a result of larger WG discussion to simplify
>operations.
>
>I think there is a larger data consistency issue that is introduced if the
>client sponsorship model is abandoned.  As changes are made across clients,
>all of the previously involved clients will probably like to know about
>changes made through other clients so that whatever locally-copied data they
>maintain remains consistent with that of the server.  It's a many-to-one
>relationship.  With the client sponsorship model, there's a one-to-one
>relationship that is _far_ easier to manage.
>
><Scott/>
>