[Ietf-not43] extensibility

[Eric A. Hall]

on 7/28/2003 12:29 PM Andrew Newton wrote:

> Eric A. Hall wrote:
> 
>>I would like to review the criteria behind this analysis. For FIRS, the
>>requirement is (1) generate an LDIF file on a daily basis, (2) upload it
>>to the server, and (3) use a server that supports the FIRS extensions.
>>That is pretty cheap.
> 
> This makes implementation and operational assumptions that are 
> simplifications of how registries operate.  It also makes assumptions 
> about the data models which I know are not true for domain registries 
> and I suspect are even less true for address registries.

I should have said 'minimal requirements', which is true. It is entirely
possible to go beyond that (voluntarily), but it is not necessary.

For example, the requirements document does not mandate a read-write
system, but only allows for it, and the minimal requirements support a
read-only service as mandated.

One step beyond the minimal requirements might be to generate LDIF
changelogs every 15 minutes or so, with the adds, updates and deletes
being applied more frequently. A far-edge usage could be made which put
LDAP at the center of the registration system (generating ~XML as output,
instead of using another database and generating LDIF as output). These
are all beyond the minimal requirements however.

>>http://www.ietf.org/internet-drafts/draft-newton-ldap-whois-03.txt

This is an example of a voluntary system which goes further than the
minimal requirements, imo.

> and appendix C of
> 
>>http://www.ietf.org/internet-drafts/draft-ietf-crisp-iris-core-03.txt

This seems to discuss some specific needs of large-scale operators, which
I think were included in the "uses FIRS extensions" sub-claim (which is
not just limited to search filters, but also includes things like query
limiters). Granted, there are going to be some deployment issues for some
registry operators, but nothing beyond the kind of issues that LDAP
servers already deal with frequently (eg, clustering and replication can
take care of most stuff, and isn't technically a full step beyond the
minimum requirements listed earlier).

As far as that goes, doing this work for LDAP is going to be cheaper than
doing this same necessary work in addition to all of the other glue
necessary for IRIS.

-- 
Eric A. Hall                                        http://www.ehsco.com/
Internet Core Protocols          http://www.oreilly.com/catalog/coreprot/