Re: the relational model of data objects *and* program objects

Bernard Peek wrote:
> No. The DNS system has a handful of root name servers and is not
> decentralised. There are cached copies of their data elsewhere but the
> slave servers defer to the root servers which have definitive copies of
> the data.

Are you sure that's right?

As I understand it, DNS has a hierarchical namespace and any node in the tree may be authoritative for the branch of the tree underneath it.

So for example if you have the domain "foo.bar.com", the root nameservers only know what the authoritative nameservers for the ".com" branch are.

You then have to go to the ".com" nameservers to find who is authoritative for the ".bar.com" branch.

And finally the bar.com namservers can tell you the IP address of "foo.bar.com".

Of course there may also be caching at any level, but there's only one place that is "authoritative" for any branch (zone?) of the tree i.e. that has the definitive answer.

(note: the exact details may not be totally accurate but I'm sure the general principle works like this)

So I'd say it is truly decentralised. Not totally decentralised like a peer-to-peer network, because you can still destroy it by chopping out the root nameservers. But what you might call partially decentralised.

This relies on the fact that domain names are a hierarchy, so I'm not sure how a similar idea would work with the standard relational model though.

Paul. Received on Tue Apr 19 2005 - 13:46:56 CEST