Introduction
Today, the rate of technological change is rapidly accelerating. Organizations are faced with increasingly difficult decisions about their choices for hardware and Software platforms. Furthermore, increasing deregulation in many countries has forced companies into more aggressive deadlines and timescales to compete and survive. At present, object oriented technology is viewed as being able to deliver more reliable and better quality software, because it enables the building of more modular software by using well-defined interfaces and hiding implementation details. This can be seen in Table P-1, which shows the results of a survey of visitors to Object World UK in 1995.
According to Leach Lea95, the top reason shown in Table P-1 was also number one in 1993 and 1994.
The total worldwide sales of object-oriented development tools by US companies in 1995 were valued at US $1.3 Billion OOS96. A breakdown of the figures shows that object-oriented languages were the largest (US $440 Million), followed by application development environments (US $380 Million) and then object- oriented databases (US $218 Million).
Object-oriented languages are quite mature; Simula (the precursor to languages such as C++) has been in existence since the late 1960s. Other languages, such as Smalltalk, have also been used for several decades. On the analysis and design side, efforts are well underway through the Object Management Group (OMG) to develop a standard modeling language to communicate object-oriented designs. The need for a standard modeling language is analogous to Electrical Engineering's need for a standard notation to communicate electrical circuit designs. A standard notation will be understood by engineers wherever they were in the world.
Object databases, however, are still relatively rarely used in production LC96. The reason for this has been that a database is often at the core of many business processes within a company. Changing it can be difficult and expensive, as there may be many application dependencies. In contrast, applications can often be re- written in another language without affecting existing applications. An organization can also switch modeling languages and notions without affecting existing systems or application designs.
The Growth of Object Database Usage
Most of the major commercial object database products were developed towards the end of the 1980s and early 1990s. Many of them were built around object-oriented programming languages and had as their primary objective providing persistence for programming language objects. The computer-aided design vendors were early adopters of this technology. Previously, their approach was typically to store design objects in a relational database or to use some proprietary or flat file system. This approach often resulted in unsatisfactory performance. To retrieve design objects, for example, many database objects may have to be read and reassembled together in memory. For storage, the design objects would need to be flattened or decomposed to be manageable in conventional record or tuple structures. Some examples to illustrate these problems have been described by Loomis Loo95. Over the years, object database technology has become increasingly popular within other industry sectors as well. Evidence of this shift towards non-engineering applications has been reported in Lea95 and shown in Table P-2.
Additional evidence showing the shift to increasing mainstream commercial usage of object databases has been reported by Barry Bar97, based on a study of twenty-four organizations. Results of another survey of 700 Information Technology (IT) professionals in DBW97 showed that nearly 50 percent of respondents indicated that object-oriented databases were a technology area that their companies may use, with the Leisure, Transport, Retail, Healthcare, and Utility industries leading the interest.
The market for this technology has grown steadily over the last decade, although it still accounts for only a few percent of total sales of database products LC96. Part of the reason for this relatively slow market penetration is that the relational vendors have not totally ignored developments in object technology, and have begun to incorporate object features in their products. This is in stark contrast to the way that the hierarchical and network vendors ignored developments in relational technology several decades ago. Relational database usage grew fairly rapidly, because there was no apparent alternative for the capabilities offered by these products. Many companies are hesitant to move to a new data management technology Hod89 partly because there is only a limited experience base to leverage. It is the aim of this book to document examples that demonstrate the use of object database products to solve real-world problems.
This book brings together, for the first time, a large collection of papers that describe first-hand experiences with object database technology. Eighteen papers are presented in four categories:
1.
Systems Architecture
2.
Applications and Design
3.
Object Database Selection and Migration
4.
Performance The papers include experiences with five of the major object database products: O2, GemStone, Objectivity/DB, ObjectStore, and VERSANT. The example applications are from a wide-range of domains, such as financial and scientific applications. We expect that this book will appeal to a wide audience, ranging from people who are just starting to look at object database technology to experience object database professionals.
References
Bar97 D. K. Barry: Just the facts, please. Distributed Object Computing. 1 (1): 56- 57, 59, 1997 . DBW97 DBWorld: Report and directory 1997/8. London: Interactive Information Services Ltd., 1997. Hod89 P. Hodges: A relational successor? Datamation. 35 (21): 47-50, 1989. LC96 M. E. S. Loomis and A. B. Chaudhri: The a to z guide to object data management (Tutorial). ACM International Conference on Object-Oriented Programming Systems, Languages, and Applications (OOPSLA'96), San Jose, California, October 1996. Lea95 E. Leach: Object Technology in the UK. Introduction to COBRA. London, UK, September 1995. Loo95 M. E. S. Loomis: Object Databases: The Essentials. Reading, Massachusetts: Addison-Wesley, 1995. OOS96 OOS: The object-oriented software development tools market. Object-Oriented Strategies, May 1996.
Table P-1 Main Reasons for Moving to Object Technology Lea95 Flexibility to change
40% Reduced time to market
24% Distributed application requirements
23% Programmer productivity
19% Application complexity
18% Ease of use
9% Financial savings
7% Don't know
5% Other
17%
Table P-2 Use of Object Databases Commercial applications
17% Other applications
14% Multimedia applications
12% Document management
9% Workflow and financial modeling
9% Mapping and GIS
8% CASE
7% Network management
5% Scientific applications
5% Manufacturing
4% CAD
1% Transportation
1% No applications will run on ODBMS
28%
Akmal Chaudhri is editor of the Open Source Projects area of IBM's DeveloperWorks Web site. He has been a regular presenter on Java, XML, and databases at a number of conferences, and is the editor of three previous books, including Object Databases in Practice, Prentice-Hall, 1998. Awais Rashid teaches courses on database technology at Lancaster University. He was previously a researcher at the Xerox Research Center Europe. He writes regularly for technical journals, and serves regularly on the program committees for conferences. Roberto Zicari is a full professor of computer science at Goethe University in Germany, and is the co-editor of three previous books.