Introduction - Database Processing: Fundamentals, Design, and Implementation - page 24

However, in addition to being able to run on many different types of machines, Oracle
Database had, and continues to have, an elegant and efficient internal design. You will learn
aspects of that design in the concurrency-control section in Chapter 10B. That excellent
design, together with hard-driving and successful sales and marketing, has pushed Oracle
Database to the top of the DBMS market.
Meanwhile, Gordon Moore and others were hard at work at Intel. By the early 1980s, per-
sonal computers were prevalent, and DBMS products were developed for them. Developers
of microcomputer DBMS products saw the advantages of the relational model and developed
their products around it. dBase was the most successful of the early products, but another
product, R:base, was the first to implement true relational algebra and other operations on the
PC. Later, another relational DBMS product named Paradox was developed for personal com-
puters. Eventually, Paradox was acquired by Borland.
Alas, it all came to an end when Microsoft entered the picture. Microsoft released
Microsoft Access in 1991 and priced it at $99. No other PC DBMS vendor could survive at that
price point. Microsoft Access killed R:base and Paradox, and then Microsoft bought a dBase
“work-alike” product called FoxPro and used it to eliminate dBase. Microsoft has now stopped
upgrading Microsoft FoxPro, now named Microsoft Visual FoxPro, but Microsoft will continue
to support it until 2015 (see http://en.wikipedia.org/wiki/Visual_FoxPro ).
Thus, Microsoft Access is the only major survivor of that bloodbath of PC DBMS
products. Today, the main challenge to Microsoft Access actually comes from the Apache
Software Foundation and the open source software development community, who have
taken over development of OpenOffice.org, a downloadable suite of free software products
that includes the personal database OpenOffice.org Base (see www.openoffice.org ) and its
sister product LibreOffice (see www.libreoffice.org ) . LibreOffice is a related development
of OpenOffice that was started when Oracle Corporation acquired Sun Microsystems in
early 2010.
Post-Relational Developments
In the mid-1980s, object-oriented programming (OOP) emerged, and its advantages over
traditional structured programming were quickly recognized. By 1990, some vendors had
developed object-oriented DBMS (OODBMS or ODBMS) products. These products were de-
signed to make it easy to store the data encapsulated in OOP objects. Several special-purpose
OODBMS products were developed, and Oracle added OOP constructs to Oracle to enable the
creation of a hybrid called an object-relational DBMS .
OODBMS never caught on, and today that category of DBMS products is fading away.
There were two reasons for their lack of acceptance. First, using an OODBMS required that
the relational data be converted from relational format to object-oriented format. By the time
OODBMS emerged, billions upon billions of bytes of data were stored in relational format in
organizational databases. No company was willing to undergo the expensive travail of convert-
ing those databases to be able to use the new OODBMS.
Second, object-oriented databases had no substantial advantage over relational data-
bases for most commercial database processing. As you will see in the next chapter, SQL is not
object oriented. But it works, and thousands of developers have created programs that use it.
Without a demonstrable advantage over relational databases, no organization was willing to
take on the task of converting its data to OODBMS format.
Meanwhile, the Internet took off. By the mid-1990s, it was clear that the Internet was
one of the most important phenomena in history. It changed, forever, the ways that cus-
tomers and businesses relate to each other. Early Web sites were nothing more than online
brochures, but within a few years, dynamic Web sites that involved querying and processing
databases began to appear.
However, one substantial problem existed. HTTP is a stateless protocol; a server receives
a request from a user, processes the request, and then forgets about the user and the request.
Many database interactions are multistage. A customer views products, adds one or more to
a shopping cart, views more products, adds more to the shopping cart, and eventually checks
out. A stateless protocol cannot be used for such applications.