Re: Shared data base access by registries

["Rick H. Wesson" <wessorh@ar.com>]

On Nov 18, 10:33pm, Dave Crocker wrote:
> Subject: Shared data base access by registries
>
> [ text/plain
>   Encoded with "quoted-printable" ] :
At 10:02 PM -0800 11/18/96, George Michaelson wrote:
> >  >_ALL_ top level domains _MUST_ be shared
> >
> >  	what are the public/open, established, mature technical standards
> >  that are to be used to provide a basis for making such a requirement?
> >
> >I think an argument might run:
>
> 	I can see that I was too cryptic.  Let me restate my question:
>
> Given the goal of having multiple registries share responsibility for
> assigning entries into the same TLD, then what standard technology will be
> used to accomplish the shared data base access that is required to
> accomplish this?

I like using CORBA it is a real cool spec, and there is a (kind of) public
domain implementition of it too.

> My question is technical, not social or political.
>
> My premise is that for us to make such sharing a requirement, we need to
> have the relevant technical standards fully in place, fully deployed and
> fully mature so that we can KNOW that we can rely on their filling our
> needs.

This is another reason I like CORBA because it is a spec from the OMG
and not something we are dreaming up here today. CORBA is matureing and
the IIOP protocol make it wonderful for internet realted objects.

> And to respond to David Conrad's line of thinking, let me stress that I do
> not, for a moment, doubt the existance of adequate technology.  Proprietary
> versions abound.  But I believe that a public service such as the
> Internet's DNS must limit itself to use of standards, for any activity
> which is imposed or otherwise required.
>
> David's response shows a 6-step process that looks reasonable -- because it
> is.  But note that it refers to a formatted message.  What's the format?
> What is the sequence of messages between requestor and responder?  All of
> this, of course, is a protocol.  Now, I LIKE trivial protocols, but the
> fact is that we don't have one in place TODAY to do this job.  The general
> indication is that the DNS dynamic update/registration mechanism will
> suffice, but it's not ready for prime time, yet.

I beleieve that it would be sufficient to describe the interfaces
to these objects and the methods that operat on them in an interface
description langauge like IDL.

I have already sone alot of this for domain templates and the such. Infact
I've written just about everything you need to run a domkain regestry using
corba in C++ with java integration too. I even hacked bind so that it
would load via a CORBA methods, and all the DNS files were stored
within an OODB.

> 	Now there IS a follow-on question, if we cannot find an adequate
> set of standards in place now (and I believe we cannot, although they are
> coming down the pike and will probably be sufficiently mature in 1-2
> years).  That question is:  can we encourage COOPERATIVE sharing?
>

I believe you can, we have just not talked about all of it enough yet.

-Rick

-- 
Rick H. Wesson