[ppml] Directory Services - Take 2

[Michael.Dillon at btradianz.com]

>     > What percentage of requests for data are CD-ROM based?
> 
> Presumably a very small number, which would seem to make it inefficient 
to 
> keep a mechanism in place to serve a small number of potential requests 
> which could just as well be served via the normal mechanism.

Unless ARIN charges a fee to recover the costs involved. In that
case, the policy should not mention CD-ROMs at all, it should
just have something about ARIN being open to send the data on
some media if the requestor covers the costs. If the requestor
offers to buy a Blue-Ray writer for ARIN in order to receive
Blue-Ray disks, then I would prefer that ARIN staff deal with
such requests without the hinderance of poorly thought out
policies.

ARIN has a woeful history of getting stuck in a technical
rut with the whois directory and we have an opportunity
here to correct that. Let's not create new ruts for the
future. Deprecate rwhois, document the existing REST
service (web form query), add an XML encapsulation of
the data (borrowed from IRIS) so that whois and REST
users can optionally request the easy-to-parse data, and
set up an LDAP server using the same data schema as IRIS.

That way we move ahead.

We get rid of rwhois which is a bad protocol and a single
bad implementation.

We fix whois by making the data parseable but we do so
in a backwards compatible manner (-x option).

We provide data in a parseable format (XML or LDAP) and
the XML format is integrated into the two existing delivery
mechanisms, whois and REST. This sets the stage for possible
future transition to IRIS.

Or, if IRIS fails or we decide that it is basicall a domain
registrar protocol and overkill for ARIN, we could either
transition to LDAP, or stick with the improved REST or
the improved whois.

Why so many choices? Simply because we are 90% of the way
there already, so it is not a lot of effort to put all of
this in place. And since none of us are crystal ball readers,
it seems wise to create a level playing field of 3 protocols
that are now roughly level in features, and see where that
takes us. 

--Michael Dillon