From: Lucas Kreger-Stickles (lucasks@cs.washington.edu)
Date: Mon Apr 26 2004 - 09:49:07 PDT
*Review: Of Objects and Databases: A Decade of Turmoil**
Carey and Dewit, Review by: Lucas Kreger-Stickles*
*
*
In their paper, the authors provide a survey of the effect that
objects had on databases (and that databases had on programming
languages) over the course of a decade (1986-1996) in a non-scholarly,
admittedly biased, way. They divide their paper into several sections
which review the state of objects and databases in 1986, examines the
current state of the art (as of 1996), and suggests analysis as to where
databases and objects are headed.
*The Past... 1986
*The authors indicate that in 1986 there was a lot of excitement
surrounding objects and at the same time many database researchers were
exploring ways to expand traditional relational databases for new
applications. Out of this grew two camps of researchers who wanted to
explore radical change to traditional databases. In one camp,
researchers thought that by combining aspects of objects with relational
databases they could create a 'one-size-fits-all' database system that
would overcome percieved limitations of traditional relational
databases. In the other camp there were researchers who envisioned
toolkits that would help users design aplication specific databases.
At the same time there were some researchers who were exploring ways
in which object-oriented languages could be extended to allow them to
have certain key database features (such as persistance and atomicicity)
so as to allow the programing languages to form the basis for
datamanagment.
*Present Day... 1996
*The authors indicate that both persistant programming languages and
database toolkits has died out, determined to be dead-ends. In
addition, the authors indicate that object-oriented databases have
failed to live up to comercial expectations. Finally, the authors
indicate that extended databases from the 80's have been renamed
Object-realtional databases, and the authors indicate that they consider
them to appear to be the likely winner. In addition, the authors
present analysis as to why certain ideas failied while others took root.
The authors also indicate that one new avenue of reserach as
object-opriented databse wrappers, programs which take legacy relational
databases and construct C++ code which has objects which represent the
data held in the databse so as to ease the process of commercial C++
development for data centric applications.
*The Future... 2006
*The authors indicate that they would like to see databases which
offer:
*A fully intigrated solution: meaning that object-relational
databases would provide full support for ADTs including inheretence
*Better object-based querries.
*Parallelism.
*Acess to legacy data-sources
*Uniform Strandards
In addition, the authors note that the above list is not so much a
predication for the future as it is their wish list for the future of
databases.
*
Thoughts
*As someone who has no background in databases the survey was easy
to read and informative. When starting to think about a new field it is
nice to know where it has already been and what about the process
succeeded and what failed. (In my opinion a core benefit of open
research is the increased efficiency gained when researchers learn from
each other and minimize redundant exploration of fruitless research)
I thought that the authors did a good job in outlining their biases
and the limits of the paper and in general I found it an informative and
easy read.
This archive was generated by hypermail 2.1.6 : Mon Apr 26 2004 - 09:49:13 PDT