Dr. Brown asserts databases are dead, dead as a dodo [was Re: First Impressions on Using Alphora's Dataphor]

From: D Guntermann <guntermann_at_hotmail.com>
Date: Thu, 2 Sep 2004 04:25:02 GMT
Message-ID: <I3EC9r.G28_at_news.boeing.com>

"Paul G. Brown" <paul_geoffrey_brown_at_yahoo.com> wrote in message news:57da7b56.0408310834.382e95e5_at_posting.google.com... > "D Guntermann" <guntermann_at_hotmail.com> wrote in message news:<I3Aqo2.817_at_news.boeing.com>...
> > "Paul G. Brown" <paul_geoffrey_brown_at_yahoo.com> wrote in message
> > news:57da7b56.0408301732.c200bd5_at_posting.google.com...
> > > lajos.nagy_at_gmail.com (Josh Hewitt) wrote in message
> > news:<1c92edeb.0408260910.60242a59_at_posting.google.com>...
[snip]

> > > Forget databases, guys. They're dead. Dead as the dodo.

> >

> > Hi Dr. Brown:

> >
> > I certainly respect your knowledge and opinion, but I'm not sure I agree
to
> > the degree you state databases are dead.

>
>   First, I ain't a Dr. And second, I'm not sure I respect my own knowledge
>   and opinions, so I'm wary when someone else says they do.
>

Well....what can I say? Sorry to hear that.

> I'd respond by pointing out that it is revolution, not evolution, that > has characterized the development of ideas in computer science.

This is a theory usenet group. Though some might disagree, the truly revolutionary theory came from the likes of Babbage, Turing, von Nuemann, Claude Shannon, E. Codd, and others. I haven't seen the same degree of clarity and depth in theory recently at all. In fact, it seems that most researchers make the mistake of coupling theory with technology products and implementations, or they scurry in group fashion to "research" and add incremental knowledge to another's maybe decent idea.

And

>    specifically, repeated revolutions in our conceptualization of what
>    constitutes a 'system' or 'program'. Vannevar Bush had dreamed up
>    hypertext and something like the HTML/HTTP based web in 1945, but along
>    the way we've endured mainframes, two-tier, three tier,
>    networked/distributed architectures, the Web and peer-to-peer.

"endured"? These "architectures" and "systems" you describe are natural consequences of incremental and evolutionary advances in hardware production technologies. Processing power, disk, memory, and interconnect technologies have all become faster and cheaper. There is no theory behind that. Cache coherence, multi-processor/multi-computer, grid architectures, protocols, and the like might qualify, but they really don't stand up to the orthoganality test. They all change as technology changes.

>

> > Could you give more concrete examples of what upcoming technology
> > theories/implementations are leading to the impending raising of the
> > independent DBMS tombstone? Thanks.

>
>     The DBMS tombstone has engraved upon it "6%", which is the growth rate
>     in license sales over the last few years. This says to people with
>     money that investment in DBMS technology will not generate
>     a reasonable rate of return. (By DBMS, I mean software products that
>     manage a shared, centralized repository or 'data bank'). The market
for
>     these products has become commoditized beneath a rigid standard and
>     any attempt to break out of that standard is met with contempt by
>     customers. Want to compete in the DBMS business? Price, price, price.

So chase the "next great thing" and follow the investment money. Good for you. You could have caught "the next great wave" ;-) with ODBMS's ten years ago.

 I interpret this metric (of 6%) to mean just the opposite. DBMS's work so well and are so stable, they have become commodities. Would you consider the personal computer a commodity? If you do, has it stymied innovation, competition, and new features?

My original question asked for some sort of concrete example, or examples. Is it GRID? Is it something like from Jennifer Widom's research in streams? Are you thinking in terms of the difference between memory and storage media (orthogonal in that both are storage) What?

>
>     On the other hand, the theory and practice of information management
>     continues to be important and to grow new limbs.

I agree in one sense. I still see the same fundamental problems with data definition and management within the context of a single system, or limited distributed systems. You seem to be saying that these problems are no longer relevant, and that we can just skip these because now we will increase scope. But what about the updateability of views and other unsolved problems? Where do they stand now?

How are we going to
> cope with inputs from a pervasive computing infrastructure? What is the

>     most useful level of abstraction at which to view distributed hardware
>     resources?

Sheth and Larson as well as others have asked these questions, framed the trade-offs, and proposed solutions (and in a more theoretical sense I might add). So what is new?

These challenges are driving requirements for which the > DBMS model of the world--either SQL DBMS or TR DBMS--is entirely unsuited.
>
Except that SQL is one of the strongest standards that I've encountered. It's universal and accepted. That is its one great strength.

>     Now I'm convinced that, because it supports declarative programming,
>     relational thinking is at least extremely useful (in my bolder moments
>     I'd even sign up for 'central' or 'necessary'). But sitting around
>     comparing the merits and demerits of Dataphor and SQL DBMS
technologies
>     seems to be a gigantic waste of time. Like squabbling about whether
Greek
>     or Latin is the true language of religious consciousness.

I agree with you.

>
>     (Why am I bothering, then? Because I'd like to be convinced otherwise.
>      It would soothe my worries about my career choices no end.)

If your objective is to find a "growth technology", then I can't help you there; but there is something to be said about theory and knowledge for theory and knowledge's sake.

Regards,

Dan G. Received on Thu Sep 02 2004 - 06:25:02 CEST