Re: Content Based Addressing

From: x <x_at_not-exists.org>
Date: Wed, 17 May 2006 11:58:46 +0300
Message-ID: <e4eofr$r6r$1_at_emma.aioe.org>


"Marshall" <marshall.spight_at_gmail.com> wrote in message news:1147831779.205371.76230_at_i40g2000cwc.googlegroups.com...
> David Cressey wrote:
> > The first time I ever saw an index to a body of data (a file of flat
> > records, to be specific), I thought it was a kludge. I figured this
sort
> > of thing would last a few years until somebody built a cheap, fast, and
> > ample associative memory. Here it is, 30 years later, and indexes are
> > still with us.

> Some of this is due to the Tyranny of C. Programming languages
> have a strong impact on processor architectures, and C has had
> more impact than probably anything else ever. C is all about
> pointer arithmetic and direct manipulation of linear buffers. And C,
> as we know, represents the ultimate in *not* separating the physical
> and the logical. If anything, the C programmer's battle cry is a
> demand for more direct exposure of the physical layer.

I see nothing wrong with indirect exposure of the power of the physical layer.
I wonder how much cost that cheap, fast and ample associative memory and where we can find it.
I wonder how can one break the vicious circle of pricing.

> > Anyways, I'm starting this thread with the idea of discussing content
based
> > addressing. Basically, content based addressing says, "I don't know
where
> > it is, but when you find it, this is what it's going to look like".
This
> > description is intentionally vague.
> >
> > I want it to cover search engines that invert some body of text, as
well as
> > indexes that permit keyed access to certain rows in a table.
> >
> > The whole idea of content based addressing seems to me to be such a
powerful
> > idea that it keeps popping up in IT all over the place. Of course, in
> > c.d.t. the RDM is going to be the first thing most people think of when
> > they ponder content based addressing.

> My candidate for next-best-place for CBA: communications. Imagine
> a communication layer that used durable subscriptions with CBA.
> Communication channels could use the same type system as
> the relational engine, and in fact the two could be integrated, so
> you could, for example, subscribe to inserts into a given table.

> It makes me drool just thinking about it.

There are products that do just that.

> > Anyways, I think that content based addressing is a large part of why
so
> > many people have used RDM and/or SQL to good advantage in making
flexible
> > use of data.

> I Agree!

And they will use XML for exactly the same reasons. :-) Received on Wed May 17 2006 - 10:58:46 CEST

Original text of this message