Re: [ietf-provreg] new epp commands?

[Andrew Newton <andy@hxr.us>]

On Jan 19, 2010, at 11:34 AM, Patrik Fältström wrote:

> On 19 jan 2010, at 16.29, Andrew Sullivan wrote:
> 
>> On Tue, Jan 19, 2010 at 04:21:07PM +0100, Patrik Fältström wrote:
>>> 
>>> IRIS solves a different problem. If you want to sync two databases with epp interface, you definitely want to use epp, as such things like roid, relationship between records etc is absolutely crucial.

Yes, IRIS was focused on end-user to registry/registrar and not registrar to registry.  That was the big difference.

>> 
>> I thought there was a way to get every single thing you could get in
>> EPP via IRIS.  No?  (I'm prepared to be wrong.  Afilias was never
>> really interested in the IRIS stuff, so I read the documents &
>> commented a little but didn't put a lot of effort into it.)

There was an IRIS serialization format for copying datasets, not necessarily synching.  Think LDIF for IRIS.

> Yes, of course you _can_, but it is much easier if the database structure etc is exactly the same for every record you want to work with.
> 

I would agree with this, except this entire discussion seems to center around the fact the everyone does something in a different way.  Which kinda leads one to question a standard here.

As an aside, the ECRIT working group also has a synchronization format for their problem space and even their simpler cases seem difficult.

>> I wonder, however, whether it's even a good idea to sync two
>> databases, especially via EPP.  My time as a database geek taught me
>> nothing so much as that having two authoritative sources for data is a
>> way to be sure that one of them is wrong.  
> 
> Well, that is how a thick registry works. By having epp as a synchronization protocol. If the registry is to have the information about who is the holder of each domain, while the registrar is to invoice the customer, the registrar definitely also will have information about the same holder. Transfer of domains and contacts, and updates of contact etc information ends up being an "interesting" game where both parties are moving around the pieces on the chess board at the same time.
> 
> In this case, we more or less only talk about the ability for the registrar to issue a query like:
> 
> - What domains am I sponsor for?
> 
> - What contacts am I sponsor for?
> 
> Etc.


Andrew is right that two authoritative sources makes the job much more difficult.  If the problem were constrained to "the registry is always right", then the task gets much easier.

As for doing this in IRIS or EPP or something else?  Resources are resources and I don't think there is much of a difference when it comes to CPU cycles, so it may as well be done inside EPP to lessen the pain of adoption and interoperability.  It seems to me that if some registries don't want to implement it, then they shouldn't.

-andy
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
List run by majordomo software.  For (Un-)subscription and similar details
send "help" to ietf-provreg-request@cafax.se