<<<
Chronological Index
>>> <<<
Thread Index
>>>
Comments on Transition to Delegation
- To: gtld-transition@xxxxxxxxx
- Subject: Comments on Transition to Delegation
- From: Eric Brunner-Williams <ebw@xxxxxxxxxxxxxxxxxxxx>
- Date: Sat, 06 Dec 2008 17:23:04 -0500
Module 5 Transition to Delegation
Sec. 5.1 Registry Agreement
Sec. 5.2 Pre-Delegation Testing
Q1. "Following completion of the Board review, ..."
Is the word "Board" correct in the initial sentence of this section?
Sec 5.2.1 Technical Testing
Q2. "operate the gTLD in a stable and secure manner."
None of the nine questions address the stability of the registry
operation, thought the system monitoring question (#7) does address
existence of data necessary to measure variation.
None of the nine questions address the security of the registry
operation, though the escrow question (#6) does address existance of
data necessary to recover in the event of operational failure.
Provisioning (#3) and performance (#9) do not necessarily provide system
stability or system integrity.
Is "stable and secure" ment in the usual sense for complex, distributed
systems designers and users, or is it a rhetorical nod to the language
of the JPA?
Q3. IDN (variant) tables (Question #1)
I'll defer this question to a subsequent comment.
Q4. DNSSEC (Question #2)
"If DNSSEC is offered as part of registry services at time of
application, can applicant comply with requirements?"
This appears to significantly understate the complexity of any
meaningful provisioning of DNSSEC as a regsitry service, though it is
consistent with the minimal act of signing a zone.
Is "signing the zone" a "registry service"? If so, are similar acts of
the zone administrators "registry services"? Will the operational
activities of key generation, key storage, key rollover and related
policies be considered distinct "registry services", and subject to
cumbersome negotiation?
Q5. Architecture load requirements (Question #3)
"... aspects of this self-certification documentation can be audited
on-site at the services delivery point of the registry."
First, are the documentation the target of the audit? This seems
unlikely, and at odds with the same phrase occuring in Questions #7 and #9.
Second, and common to this and Questions #7 and #9, what is the
"services delivery point of the registry"?
Q6. IPv6 provisioning (Question #4)
While I applaud the general sense of anticipating a transition from a
pure IPv4 address service model for A and other records, it is not the
case that IPv4 is "going away", and it is quite reasonable for
registries to restrict, and not necessarily on a v4 or v6 or both
granularity, the addresses they allow to be associated with the registry.
This is a good "pro forma" test of regsitry technical ability but it is
not a good policy to confuse v4 address exhaustion for general
allocation with v4 address availablity for resolved network resources.
Finally, what justification can be offered for asking question #4 in the
first place? Do we care if names in .foo can only be on v4 addresses? Is
there anything wrong with that as a business model? Is there anything
right with the converse, if names in .bar can only be on v6 addresses?
(There's lots of things wrong with that.)
Q7. IPv6 publication (Question #5)
Again, good general sentiment, but is v4-only service a bad thing for
all registries? Suppose only v4 addresses are available to a bounded
region, and the entire community served by a registry was within that
v4-only region, what is the point of asking for v6 on the publication side?
Finally, for both Q6 and Q7, the use of IPv6 is somewhere between 0.1%
and 0.2% of the global access address space, with a high of less than
three fourths of a percent in Russia and two thirds of a percent in France.
Q8. Escrow
Comments provided separately.
Q9. Continuity
Comments provided separately.
Sec 5.4 Ongoing Operations
Q10. Given the number of applicants who have no prior experience with
either ICANN's gTLD registry liaison (or even the regsitrar liaison), an
enumeration of the "assistance" available to present applicants as
future operators would be a useful clarification. The "assistance" can't
(usefully) mean compliance auditing.
<<<
Chronological Index
>>> <<<
Thread Index
>>>
|