25th International Conference on
Very Large Databases
Edinburgh - Scotland - UK
7th - 10th September 1999

VLDB Success Stories

The International Conference on Very Large Data Bases (VLDB) is primarily an event where academics describe their research work. It is, however, the premier conference on Database research and much of the work which is reported at the conference eventually finds its way into commercial products. Here are three success stories where the link between presentations at VLDB and commercial products is clear.

Postgres and DataBlades

The 1987 VLDB conference in Brighton included two papers by Prof. Mike Stonebraker of the University of California at Berkeley entitled: "The Postgres Data Model" and "The design of the Postgres storage system". Prof. Stonebraker was already well known in both commercial and academic circles for the design of the Ingres relational data management system. Ingres grew out of academic research at Berkeley and became an industry strength product marketed by the Ingres corporation. The Postgres project was a piece of academic research that sought to extend the existing Ingres database. Its name derived from the fact that it was post Ingres.

Postgres pioneered the concept of an object-relational database management system. It used relational data management techniques but was also able to cope with complex objects such as pictures, sound and video. The key technique it introduced was the ability to incorporate new types of data into proven relational technology.

In August 1992, Prof. Stonebraker co-founded Illustra Information Technologies, Inc. in order to commercialise his research in object-relational database technology. The most important aspect of the Illustra product was taken straight from Postgres: the ability to handle complex data objects. This ability was given the marketing name: DataBlade.

On February 16th 1996 Informix, the world's second largest relational database supplier to the UNIX marketplace announced:

Informix Corporation, the leading provider of parallel processing database technology, today announced that it has completed the acquisition of Illustra Information Technologies, Inc. Illustra is a leading supplier of dynamic content management database software and tools for managing complex data in the Internet, multimedia/entertainment, financial services, earth sciences, and other markets. With the acquisition, Illustra becomes a wholly-owned subsidiary of Informix Corporation.

Informix will issue approximately 12.6 million shares of Informix common stock to acquire all outstanding shares of Illustra. An additional 2.4 million Informix shares have been reserved for issuance in connection with the assumption of Illustra's outstanding stock options. Informix currently has approximately 135 million shares of stock outstanding. The transaction was accounted for as a tax free pooling of interests.

The acquisition will enable Informix to integrate Illustra's dynamic content management system, including existing DataBlade(R) modules, into its core parallel database technology, Dynamic Scalable Architecture (DSA). The combined technologies will result in the INFORMIX-Universal Server scheduled to ship in the fourth quarter of 1996.

The academic research described in detail at VLDB 1987 had over the following ten years evolved into a technology worth millions of dollars.

The original papers presented at Brighton are now considered classic texts and have been reprinted in many collections of relational database writings. The proceedings of the Brighton conference have been sold out for some time and further copies are no longer available.

Infomodeler a Conceptual Schema Design Tool

In 1989 VLDB was held in Amsterdam. One of the tutorials given there was on the topic of Conceptual Schema design. This was delivered by Prof. Shir Nijssen and outlined a conceptual modelling technique known as NIAM (Natural language Information Analysis Method). This coincided with the release of a textbook on the same topic, "Conceptual Schema and Relational Database Design" co-authored by Prof. Nijssen and Dr. Terry Halpin.

Terry Halpin continued to develop the NIAM methodology into what is now known as Object Role Modelling (ORM). He also worked on methods for automatically generating relational database schemas from ORM designs. The results of this work were published in a series of research papers.

One of the outcomes of this research was a tool for database design. This was adopted by the database division of Asymetrix and christened 'InfoModeler'. Terry Halpin took leave from his University post to head database research at Asymetrix.

In 1996 the database division of Asymetrix became a company in its own right called Infomodelers, Inc. It is currently selling version 3.1 of the InfoModeler product. InfoModelers, Inc. say on their web site:

"InfoModelers is a leading supplier of state-of-the-art visual design tools for database developers and users based on both conceptual and logical modeling technologies. InfoModelers, Inc. began as a division of Asymetrix Corporation and was spun off as a separate company in October 1996 to pioneer the development of database tools incorporating both Object Role Modeling (ORM) and Entity Relationship modeling (ER). ORM uses natural English sentences to express relationships between objects and simplifies modeling by allowing database designers and business users to communicate directly and easily.

Privately held and a member of Paul Allen's Wired World portfolio, InfoModelers offers the world's only database modeling and query software incorporating the innovative Object Role Modeling (ORM) methodology."

"InfoModelers provides data modeling tools that deliver the right methodology for database developers and administrators. InfoModelers' state-of-the-art visual design tools are based on Object Role Modeling (ORM), a conceptual information modeling approach that lets the user build correctly normalized data models using English facts and examples. InfoModelers' products allow businesses to focus on solutions, not database theory.

Because ORM is based on simple language sentence structures, users of InfoModeler have a better context from which to model data. Modeling in this manner is a more reliable way to accurately capture business objects, rules, and constraints while building database designs. These conceptual models are mapped to logical models that then map to the leading relational database products on the market today."

There are over 7000 companies world-wide using ORM. Major users include: 3M/Imation, Abbott Laboratories, American Express, Andersen Consulting, AT&T, Bank of America, Bankers Trust, Boehringer Mannheim, Boeing, Bonneville Power, Cap Gemini, Deloitte & Touche, EDS, Fidelity Investments, Hewlett-Packard, Intel, Microsoft, Nordstrom, Parke-Davis, Price Waterhouse, Sandia National Laboratories, Unisys, US Air Force.

In May 1997, Infomodelers and Informix announced a strategic partnership. With the release of InfoModeler 3.0, InfoModelers became the first data modeling company to provide extensive support for the new features of INFORMIX-Universal Server.

As with DataBlades, this is a technique which emerged from academic database research and was featured in an early stage at a VLDB conference. Since then it has become a major commercial success.

O2: an Object-Oriented Database

At the same VLDB (Amsterdam 1989) which saw the beginnings of Object Role Modelling, two papers were presented on a research system called O2. These were entitled "The O2 Object Manager: an overview" by Velez, Bernard and Darnis and "The O2 Database Programming Language" by Lécluse and Richard.

The papers described the results of the three years activity within the Altair research consortium. This work was part of a five year project dedicated to the development of an advanced database management system.

In 1991 when the project came to an end a company called O2 Technology was set up to market, support and develop the O2 system.

O2 Technology describes its role in the Object database market as follows:

"O2 Technology has been one of the main actors of the definition of the object database standard: the ODMG-93. O2 Technology is currently shipping the first object database compliant with this standard.

The Object Database Management Group (ODMG) was created in 1991. The initial version of the standard was produced by five object database vendors (O2 Technology, Object Design, Objectivity, Ontos and Versant ) under the chairmanship of Rick Catell. This standard is a major event and a clear signal for the object database market, as important for object databases as SQL was for relational databases.

The ODMG standard is a 'portability' standard, it guarantees that a compliant application written on top of compliant system X can be easily 'ported' on the top of compliant system Y. This is opposed to an interoperability standard, such as CORBA, that allows an application to run on top of two compliant systems (X and Y) at the same time. Portability was chosen because it is the first requirement of the users.

Three interfaces have been defined: ODL (Object Definition Language), OQL (Object Query Language) and connections to C++ and SmallTalk.

The object database market represented more than $100 million in 1995. The market is growing exponentially and its development is greatly accelerated by the generalization of client/server architecture, the standardization of graphic interfaces, the emergence of multimedia applications, and the acceptance of the ODMG-93 Standard.

Object database systems bring developers a full range of advantages inherent with object technology: faster development, maintainable applications and reusable code. "

O2 Technology is now ranked number one in the European object database market. It has seventy employees world-wide. The query language of the O2 database system has formed the basis of the recognised standard query language for object databases: OQL.

O2 Technology's customers include: Bellcore, BT, British Aerospace, Ford Europe, Glaxo Wellcome, Japan Atomic Energy, NASA, Toshiba, Xerox.

Again VLDB provided a forum where the details of an exciting and innovative product were made available to delegates before the product appeared in the commercial marketplace. Attendees at the talks were available to evaluate the impact of this new technology on their organisations two years before it became commercially available.