Re: Object Oriented Databases

["Sean R. Durity" <sdurity@cornercap.com>]

Like any model (even object), the relational model does well for some
types of data and less well for others. 

RDBMS have several advantages (as others have noted): speed,
scalability, wide-spread usage, time-tested, (some) portability of data
and schema, tools availability, can connect with non-object-oriented
tools, and relational SQL can be much easier than nested vectors of
objects. ODBMS seem to be a niche product or novelty. They have not
caught on, I would say, because of widely varying implementations and
lack of stability and scalability.

It gets interesting when you look at RDBMS with object technology, like
IBM Informix IDS. While it is a fast (the fastest) RDBMS, it also has
object abilities such as: User Defined Types (like an object, can be
programmed in C, Java, or SQL), table inheritance (Table B can be a
subclass of table A - receiving all of A's columns, plus it's own. The
overlapping data, is not duplicated!), Java stored procedures that can
run inside the engine's own JVM, User Defined Aggregates, and more.

On Tue, 2002-07-30 at 17:01, Stieglitz Noel (fin2nxs) wrote:
> Hi all,
> 
> Does anyone have any experience with ODBMS?  I have always been curious why
> they have not caught on.  It seems to make a lot more sense than dealing
> w/impedance mismatch, JDBC, etc.  If you want to develop a new, complex
> system, ODBMS has a lot of good qualities.
> 
> Any thoughts?
> 
> 
> Noel
> 
> 
-- 
Sean R. Durity
IT Manager/Developer
CornerCap Investment Counsel (http://www.cornercap.com)
Disciplined, long-term investors 
Private portfolios, institutional accounts, and mutual funds