Re: Anti Capitalism?

[davidk@ISI.EDU]

Perry,

> Perry E. Metzger writes :
> 
> 
> Kent Crispin writes:
> > Perry E. Metzger allegedly said:
> > > > Would
> > > > the back end repository not also be responsible for generating the
> > > > DNS db files and also running a whois service not to mention publishing
> > > > all the requests for domains?
> > > 
> > > It would be responsible for generating reports that would drive the
> > > generation of DNS db files and permit the generation of the whois
> > > database, but I doubt the machine would be running whois itself.
> > 
> > So, an obvious question is...who runs whois?
> 
> I'll be a bit more informal here than ususal -- this is really just
> off the cuff from me.
> 
> Who said whois needs to be "run"? "Whois" is very nearly a text search
> service on a bunch of contact data -- a comparatively small database,
> too. Off the cuff, I'd say, generate the database nightly and put the
> database up for FTP along with daily diffs, and let whomever wants to
> put up a server. If CORE feels like funding a server or two of their
> own, thats fine, too. I suspect the data is useful enough that lots of
> people would replicate it. Of course, long run, whois qua whois is
> probably dead, given rwhois and such.
> 
> Incidently, a machine capable of handling high rates of whois hits
> wouldn't have to be especially big or crunchy. I'm just pointing out
> that the functionality is segregable.

I don't think that the registry repository should contain the whois data
at all. The repository should be as lightweight as possible, that is:

- pointer to registrar that registered the data
- contain DNS configuration data
- public keys of registrars
- public keys of the domain holder (needed for transfer of
  domain registration services between registrars)

The number of updates to the repository will then be the minimum since
one only needs to update it in case of:

- a new domain
- a change in DNS configuration data
- a new registrar

but *not* for administrative reasons like updating contact records.

(You might be able to run this on a single high-end Linux box ;-)
 and you can use the most simple database product for this)

The repository makes a dump of this data available every day/week. Diffs
can be made available real-time (on a separate machine).

The whois services can be run by the registrars (or other parties) and
can support redirects to other registrars by making use of the repository
dump.

Please keep the centralized part as simple as possible. This will
increase reliability, performance and possibly reduce the 20k fee ;-).

Note that it is important that these requirements are mentioned in the
IAHC document to make sure that CORE cannot make it self more important
(by requiring more functionality in the repository) then strictly needed
(and thus make it more difficult for new registrars to enter the
business).

David K.
---