[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [DotGNU]Add Dynamic DNS to DotGNU

From: Jonathan P Springer
Subject: Re: [DotGNU]Add Dynamic DNS to DotGNU
Date: Thu, 2 May 2002 16:12:35 -0400
User-agent: Mutt/1.3.28i

Let's be careful to fully understand the problem before we jump into the
pool.  The issue to be dealt with here is the long-standing difference
between Naming and Addressing.

To define the terms as I use them:

+ Naming - to provide a unique identifier for some resource.

+ Addressing - to identify a physical means to access a given resource.

To date, DNS has tied naming and addressing together.  Looking up the
name "" yields the address .  Nothing new

The conceptual problem arrives from two directions URLs/URIs and standards
that have been derived for namespaces in languages like Java.

We'll tackle the last first.  Most global unique identifiers use
registered domain names.  Why?  Because the administrative processes are
conveniently in place to ensure that if you control a registered
namespace, then nobody else does.  Very convenient.  Makes a lot of
sense.  Confusing as heck, because apart from administrative convenience
the name has NOTHING to do with IP addressing.

URLs as an example of URIs/URNs (don't get me started) further confuse
matters.  Most URLs (e.g. HTTP, FTP) include a portion that is
explicitly intended to be looked up in DNS and used to physically
resolve connection issues.

OK, why the long lead-in?  I think what Seth's asking for is really a
core of DotGNU -- independence between how a service is identified (its
Name) and where it executes (its Address).  How do we accomplish that?

I'm sure others have thought of this, too.  The best analogy I see is to
how gnutella searches for named files through a P2P network.  The big
issue remaining is trust.  If I publish a service "" on my
server, what's to prevent Gopal from pulishing a malicious
"" clone on his?

The key point that I hope all will remember is that we MUST separate
service names from physical addressing.  Perhaps its obvious to others,
but it wasn't until I saw Seth's memo that I realized the danger.


On Thu, May 02, 2002 at 02:50:22PM -0400, Seth Johnson wrote:
> I think I really should suggest the following.
> I think it would be a very, very good idea for the DotGNU
> framework to include provision of dynamic domain name
> support.
> This would help spread the notion of people running their
> own servers, and help repudiate the whole idea of people
> being consumers of remote web services.  It would also
> emphasize the two-way nature of the fundamental internet
> protocols, emphasizing the communications medium nature of
> the web versus the "content delivery" notion.  It would also
> be a very important component in distributed/P2P stuff --
> perhaps including VRS.
> I would specifically facilitate adoption of the distributed
> stuff, and make a strong impression in terms of what DotGNU
> means for Mr. Everyday Joe.
> So DotGNU servers of whatever stripe, having static IPs,
> would have a standard service enabling them to host dynamic
> domain names appended to their own.  This would allow all
> sorts of internet users to set themselves up with URLs and
> enter their own computers into cyberspace, running their own
> DotGNU servers and nodes.
> Dynamic domain name registration could be built right into
> installation packages for client or node server software!
> This would add the notion of "everybody can build and run
> servers" to the more restrictive remote "content" delivery
> model of web services.
> I'm just stream-of-consciousness-ing here.  I'm sure I can
> say a whole lot more, because this notion has been bouncing
> around in my mind for a long time, but I didn't want to wait
> for the "composition cogitation" part of my brain.
> :-)
> Seth Johnson
> _______________________________________________
> Developers mailing list
> address@hidden

-Jonathan P Springer <address@hidden>
"A standard is an arbitrary solution to a recurring problem." - Joe Hazen

reply via email to

[Prev in Thread] Current Thread [Next in Thread]