Re: Pizza Example

From: Dawn M. Wolthuis <>
Date: Sat, 10 Apr 2004 07:45:14 -0500
Message-ID: <c58q94$jm1$>

"Jan Hidders" <> wrote in message news:urPdc.66253$
> Dawn M. Wolthuis wrote:
> > "Jan Hidders" <> wrote in message
> > news:oIBdc.65756$
> >>Dawn M. Wolthuis wrote:
> >>>"Jan Hidders" <> wrote in message
> >>>news:KGmdc.65126$
> >>>
> >>>> [...] Could you, just to indulge me, sketch for me what you
> >>>> think the trade-offs are and under what circumstances you would
> >>>> prefer one type of system over the other?
> >>>
> >>> That is really the big question I'm trying to answer -- why,
> >>> after knowing Oracle and other RDBMS software, would I choose to
> >>> go wtih a database that doesn't follow the rules for being a
> >>> database? I'm still trying to pinpoint that myself.
> >>
> >> So, since you have thought so very deeply about this you surely
> >> know the textbook answer to that question. Is your experience in
> >> agreement with that answer or does it contradict it?
> >
> > Well, I know several text book answers -- for example, using an RDBMS
> > makes changes to software database applications less expensive and
> > less risky. My experience tells me the opposite.


> My experience tells me such broad generalizations are almost always
> wrong. It is rather obvious that the data independence that is offered
> by RDBMSs is not absolutely always essential, and there are undoubtedly
> cases where it makes things in fact more difficult. But I believe it is
> also fairly well known for which type of organizations, applications and
> changes it *is* crucial.

> If you have made a serious effort to answer "the big question" then you
> will have checked if those typifications match with your experience. So
> did they? And if they didn't, can you explain why the common-sense
> arguments that justify them were incorrect?

The way that I state what I'm looking into are "thick of the bell curve" data-intensive business software. By that I mean that the risks are dollars and not lives; the database is not in the VLDB category; security is very important, but again does not risk lives; user interfaces are required and cannot be exclusively character-based green-screen applications; web services or at least data exchange mechanisms are important or will be in the future; a Microsoft Access or Excel solution is unsatisfactory; it is not an embedded database application and there is no assumption that a single application will use the data; and so on.

One of the categories of interest are ERP solutions for mid-size businesses: manufacturing, higher ed; health industry; service organizations. Many such companies invest in RDBMS solutions. I have anecdotes, not evidence, that backs up my intuition, not evidence, that the use of non-RDBMS solutions (including PICK, Cache', Berkeley DB, ...) over a 10 year period is significantly less expensive. I have found no surveys or experiments that shed any light on this, so if anyone knows of any ...

--dawn Received on Sat Apr 10 2004 - 14:45:14 CEST

Original text of this message