Good Morning,

I think I need to start at the beginning. Why are we wanting this solution? Is 
there a problem that we are trying to solve with this draft? Are registrars 
asking for this, it seems like a large burden for registrars?

As I have mentioned several times before, if this is a proposed solution for 
supplying the optional WHOIS display field Reseller Name from the registrar to 
the registry than I believe this solution is not appropriate, this is way too 
heavy handed and I don't see many registrars using this solution. Most of this 
information, registrars will not want to share and they will not want the 
management headache of maintaining all of this data at many different 
registries. Management of this would be a huge burden on registrars, trying to 
keep all the registries updated with this duplicated data every time some piece 
of reseller information changes.

If this internet-draft is trying to solve some other issue it would be good to 
understand that issue.

Assuming there is some other reason beyond using it as a solution for the 
optional WHOIS field of Reseller Name than I think we need to address some 
possible confusion on the definition of a Reseller. In the internet draft it 
seems fairly clear that a registrar is the customer of the registry and that 
some registrars have "agents" called resellers that may sell domain 
registrations on behalf of the registrar (section 1 of internet-draft). But the 
drawing that Linlin provided and the definition of Parent Identifier makes it 
sound like we want to call a Registrar with "agents", a Reseller, am I reading 
that correctly?


Thanks
Roger


From: regext [mailto:regext-boun...@ietf.org] On Behalf Of Gould, James
Sent: Tuesday, August 23, 2016 8:52 AM
To: Antoin Verschuren <i...@antoin.nl>
Cc: Rik Ribbers <rik.ribb...@sidn.nl>; Linlin Zhou <zhoulin...@cnnic.cn>; 
regext <regext@ietf.org>
Subject: Re: [regext] reseller drafts reviews

Antoin,

Good points.  My feedback is embedded below.

-

JG


[cid:image001.png@01D201FB.F4407610]

James Gould
Distinguished Engineer
jgo...@verisign.com

703-948-3271
12061 Bluemont Way
Reston, VA 20190

VerisignInc.com<http://VerisignInc.com>

On Aug 23, 2016, at 7:20 AM, Antoin Verschuren 
<i...@antoin.nl<mailto:i...@antoin.nl>> wrote:

Op 22 aug. 2016, om 16:02 heeft Gould, James 
<jgo...@verisign.com<mailto:jgo...@verisign.com>> het volgende geschreven:

Your response made me think a little bit more around the problem that is being 
solved.  I see 2 problems that could be solved by formally defining the 
reseller and linking the reseller with objects in the registry:


  1.  Displaying the Reseller information in RDDS (RDAP).  Right now it is the 
reseller name that needs to be displayed, but what if more information is 
needed later like E-mail, URL, Address, Contacts, etc.?
  2.  Enable registrars to set security, policy, and reporting options for 
reseller's at the registry level.  I can see many different use cases here that 
we've not yet fully discussed.

Thoughts?


Hi James,

I fully agree with your use cases. Thank you for clarifying them.

In addition this was actually my main question that has not been answered yet 
thinking ahead.
Looking from a registration/registry perspective I think that a 
registrant/registrar/reseller/dns-operator is only a role an entity has for a 
specific registration.
It is NOT a tag to the entity itself. You cannot say an entity IS a reseller, 
only that he performs a role as a reseller for a specific registration. It's 
the relationship that is tagged, not the entity.


You are correct that in reality, an entity could play multiple roles, but it's 
the roles that are relevant from a system perspective to apply security, 
policy, functional, and reporting options.  You could certainly have a 
conglomerate that serves multiple roles, but there are attributes that will be 
unique to each role.  I don't believe that the registry should not have 
multiple copies of information not applicable in the security, policy, 
functional, and reporting options defined in problem #2, but that may be 
unavoidable.  I view the role taking priority over the entity in the domain 
registration data model, where there would be a role object that could 
reference a common entity, but it does not need to.


Outside the registry relationship, an entity is appointed or "tagged" as a 
reseller by a registrar, but that contract is not what a registry administers.

What happens if an entity is a registrant for one registration, a dns-operator 
for another one, a registrar for some others, and a reseller for yet another 
group?
Will he be in the registry 4 times in 4 different tables, or only once?

Let's break down the data model based on who is authoritative for the 
information.  Let's say that "Jim" plays all of these roles for most likely a 
completely different set of objects:


  *   "Jim" the Registrant - "Jim" the Registrant would register as a 
registrant with a registrar, who would be authoritative for the information.  
"Jim" the Registrant could register many domains that get registered in the 
associated registries (e.g. jim.example1 in EXAMPLE1 registry, jim.example2 in 
EXAMPLE2 registry).  Do EXAMPLE1 and EXAMPLE2 registries really need "Jim" the 
Registrant information to fulfill their service?
  *   "Jim" the DNS Operator - I believe we would need to discuss the domain 
registration data model, what functions are available in the registry, and who 
is authoritative for what information for "Jim" the DNS Operator.
  *   "Jim" the Registrar - For ICANN accredited registrars, "Jim" the 
Registrar would be registered with ICANN, so ICANN would be authoritative for 
"Jim" the Registrar with the IANA ID as the globally unique identifier.  The 
registry would need a reference to the registrar to map up to the registry 
credentials, finances, and to link to registry objects.  For non-ICANN 
accredited registrars, then the Registry would be authoritative for "Jim" the 
Registrar unless there was some central non-ICANN accredited registrar 
authority.
  *   "Jim" the Reseller - "Jim" the Reseller has a contract with a registrar, 
who would be authoritative for the information.  "Jim" the Reseller's registrar 
could create a link to "Jim" the Reseller in the registry that could be used to 
tag registry objects.  The only information that is needed in the registry is 
the bare minimum to enable the linkage to solve problem #1 and #2.



What if he is a reseller for multiple registrars?

I would assume that he would be created as a separate entity for each 
registrar.  The relationship is between the reseller and the registrar, where 
there is no need for a central repository that all of the registrars link to.



What happens when he changes emailadress/phone number/address?

He has to change the information in each place that is authoritative for the 
information.  "Jim" the Registrant could register domains with multiple 
registrars, but there is no need for the registrars to reference a single 
instance of "Jim" the Registrant.  Data is copied in this instance to multiple 
authoritative repositories, since there is no single authoritative repository 
for all entities and all entity roles.



What happens when a reseller becomes a registrar for a new set of domains but 
remains reseller for the existing ones? Will his identifier and credentials 
change?


No, they are viewed as different entities entirely.  Yes, it may be exactly the 
same organization, but they are viewed as different entities from a system 
perspective.



In my humble thought experiment, I think we should register a reseller the same 
as we do a registrar.

The main difference between a registrar and a reseller is the contractual and 
business relationship.  The registrars are setup in the registry with 
credentials, financial information, and access based on the contractual 
relationship.  A reseller has no direct relationship with the registry, so 
although it is an organization, it has a completely different set of features 
from a registrar.   It would be up to the registrar to register their resellers 
with the registry to be able to apply the appropriate security, policy, 
functional, and reporting options at the reseller level.


As long as he does not have a registrar role for any registration, but only a 
reseller role, far less data is required, but the entity will need to get an 
identifier and credentials that do not change once he will get an additional 
role.

I don't believe that we would want or need to attempt to merge all of the roles 
into a single entity to support items like a reseller morphing into a 
registrar.  The key is to keep the authoritative data with the authoritative 
source and link to those sources from non-authoritative parties.



So it's business logic to determine how much data is required and displayed 
depending on the role the entity has in a specific registration. But a phone 
number is a phone number, an emailaddress is an emailaddress, and there is no 
difference in definition in emailaddress requirements of a reseller compared to 
that of a registrant or any other entity that needs to be administered. Once 
the local registry policy determines the emailaddress field is mandatory for an 
entity in a reseller role, the definition of an emailaddress is still the same..
Can this be done, especially looking at the way registrar and registrant 
entities are treated different now?


Should the registry be the one to provide this information?  The registry may 
know of the relationships to support functionality, but having the registry get 
a copy of the data for display and making decisions as far as what to display 
runs into data privacy issues.



The question of authoritative data versus informational is an interesting one.
We need to know why the data needs to be in the registry in the first place.

I believe that we should look at the data model for domain registration data 
and consider who is authoritative for each entity in the data model.  Pushing 
data down to the registry simply to support displaying it in RDDS is not the 
correct architecture that scales and that meets the data privacy issues.  In 
DNS we got away from the use of host files to use of a delegation resolution 
model.  RDAP supports a delegation resolution model, so why not leverage it 
instead of pushing data around?



If the data is to be used later for limited access rights to the database, like 
a dns-operator allowed to change an NS-set or a reseller allowed to change a 
registrant's emailaddress, then it needs to be authoritative data from the 
registry perspective. Do we think that will never happen, or do we leave that 
decision to the business case of each individual registrar? I can imagine some 
registrars want to delegate some specific access rights and reporting, and 
others don't. Should we provision for both, or only one use case?

The question for DNS operators as it is for any other domain-based service 
operator (web, e-mail, DNSSEC, parking, secondary market, etc.) is whether they 
should be formally defined in the domain registration data model and if so who 
would be authoritative for the information, what role does the registry play, 
and what privileges would they have.



Thoughts?

- --
Antoin Verschuren

Tweevoren 6, 5672 SB Nuenen, NL
M: +31 6 37682392


_______________________________________________
regext mailing list
regext@ietf.org
https://www.ietf.org/mailman/listinfo/regext

Reply via email to