From: Li Yan (lanti@u.washington.edu)
Date: Sun Apr 25 2004 - 22:33:07 PDT
Of Objects and Databases: A Decade of Turmoil
This is a very interesting paper, both its content and the
context we are reading it. The author is trying to give some
predictions to 2006, and we are right in 2004, a good time
to see the truth and the possible truth in object-relational
database.
The idea to support objects in database systems has been
originated long ago, before the 1986 where the authors
started their overview on this topic. Four different
approaches were pursued by different groups:
1. Exteneded relational database systems
2. Persistent programming languages
3. Object-oriented database systems
4. Database system toolkits/components
And a decade later, in 1996, the first two essentially
disappeared in major commercial effort and very litter
research progress were attempted. OODBMS has emerged and a
few start-up companies were shipping products, with
standards such as ODL, OQL available. However, it has never
reached the fast growing pace as predicted by OODB
advocates. On the other hand, the development of extending
relational database to support objects has benifited a lot
from OODB development and finally evolved to be the dominant
player in this competition, the ultimate concept of Object
Relational Database Systems. Major players in database
market today have already provided products implementing the
features specified in Object-relational database, e.g.,
Oracle has announced Object-relational as a major feature
for the coming Oracle 10g.
It is not surprised at all to see the least radical approach
won at last, given the fact that the relational database
system is so successful with well-developed products and
user group, a standard, SQL, and exclusive support from all
major DBMS vendors, moreover, it can also assimilate advance
in the other three directions, although in a somewhat
restricted way. Backward compatibility is always a concern,
and in most cases, the major concern in software design. And
here again, we see the one with a far better backward
compatibility eventually wins.
Although we are now in 2004, the last a few years in the
final year of the authors projection, I personally do not
have enough experience in the object-relational dabase
system's development. But support of user-defined ADT are
widely available, at least in Oracle and PostgresQL, row
type should be there as well, since these two share some
similarities in implementation. JDBC is becoming popular as
a competing database connection to a variety of DBMS, with
almost all major DBMS providing their version of JDBC
support. The success of Java programming language and JDBC
confirmed the authors call for help.
It is generally hard, or impossible to make prediction, even
in the near future, and we all have heard about "silly"
prediction made by well-known, smart scientists and
engineers. But the authors here took a rather conservative
way in making their projection in 2006, by describing the
commercial database development in the coming decade. Given
the latency between research development and commercial
implementation, they probably got most of their words
correct.
This archive was generated by hypermail 2.1.6 : Sun Apr 25 2004 - 22:33:09 PDT