[gnso-rds-pdp-wg] Proposed Agenda for RDS PDP WG Meeting Tuesday 25 April 16.00 UTC

[Andrew Sullivan]

Hi,

On Wed, Apr 26, 2017 at 03:16:04PM -0400, Greg Shatan wrote:
> one registrar I know of accesses approximately 10,000 WHOIS records a day.
> Slicing and dicing the data currently in WHOIS could have massive ripple
> effects (maybe tsunami effects is more appropriate), if one look-up now
> becomes several look-ups later.  The possible split between thin and thick
> is bad enough.

10,000 whois records in one day is a trivial amount of data.  The
whois record for yitter.info, for instance, is 3398 bytes; so 10,000
of those would be a little over 3M over the course of a day, or an
averate of about 393 bytes per second.  You could run a webserver that
operated at that level on your phone, and not even notice the
overhead.  

RDAP is designed precisely to distribute data widely and to allow
multiple follow-up transactions that cause additional queries; it pays
for this in part by transferring less data per request.  In this, it
operates just like every other web service on the Internet today.
There is nothing remotely unusual about this architecture, and I don't
think that this WG ought to spend any time second-guessing the
underlying technology, or attempting to make policies based on ideas
about what is or is not possible in the RDS.  It is very far from
plain to me that we have the expertise here to undertake such
evaluations, and the people who are competent to make such evaluations
already designed and delivered RDAP.

> Also, if some of the data is taken from the best possible (avoiding
> "authoritative") source (BPS), rather than the RDS database being the best
> possible source, then that should be explained and understood.

I think that's worrying about operational matters that are mostly just
configuration.  This group should set some policies and permit the
operators to figure out how to implement them, as long as we don't
attempt to constrain operations to unreasonable levels of service.  (I
think a requirement for an RDS that is guaranteed synched with the
registry/repository database would be too high; I think a requirement
that the RDS normally not lag by more than (say) an hour would be well
within the bounds of operational reason.)

Best regards,

A

-- 
Andrew Sullivan
ajs at anvilwalrusden.com