French OODBMS newcomer strong at starting gate; O2 Technology seeks differentiation to extend popularity beyond France - object-oriented database management system vendor O2 Technology - Global Software Experiences - Company Profile

Many in the European research and development community lament that not enough of their innovations make it to market. Some suggest this is because the R&D community is not interested in the banality of selling.

"The French have always been very good at developing great technological innovations, but poor at marketing them," maintained Francois Bancilhon, chief executive officer for Versailles-based O2 Technology. The firm is one of the newest players in the object-oriented database management system (OODBMS) market.

"It's a whole different game, listening to customers and shifting your interest from striving for a technically perfect product to trying to make a commercial success," explained Bancilhon, former head of the French-led Altair research consortium, which spun off O2.

While a little early to attribute commercial success to O2 since the fledgling company's technology just began shipping last June, some 35 French clients have judged it to have commercial potential. Included in that list are the French utility Electricite et Gaz de France (EDF/GDF); the Paris-based rapid transit agency, RATP; the national railroad, SNCF; Aerospatiale; and the Commisariat d'Energie Atomique (CEA).

O2 Technology is part of the third generation of OODBMS vendors--those vendors which market products that have a substantial amount of research behind them, said Judith Jeffcoate, associate consultant at Ovum Ltd., a London-based market research firm. The Orion OODBMS, which came out of the U.S. Microelectronics Computer Technology Corp. (MCC) consortium, based in Austin, Texas, is another such product, now being being marketed by Itasca Systems, Minneapolis.

The five years of research behind O2's technology was initiated in 1986 by Inria, France's national institute for research in informatics and automation, based in Rocquencourt, near Paris. Inria set up the Altair consortium, which comprises Inria; Siemens-Nixdorf, Munich, Germany; Groupe Bull, Paris; the University of Paris XI; and CNRS, the national center for scientific research, Paris, to design and implement a next-generation DBMS.

After investing 130 man-years of development effort, the final O2 system was completed and tested internally in September 1990 and went into beta test three months later. The company was formed in mid-1991 with Bancilhon leading a number of former members of the consortium.

The research invested in O2 contrasts sharply with some of its American competitors, such as Objectstore from Object Design, Inc., Burlington, Mass., and Objectivity from Objectivity, Inc., Menlo Park, Calif., which was brought to market after eight to 12 months of R&D. Thos companies, said Jeffcoate, have followed a more practical approach to OODBMS technology by bringing their products to market more quickly.

Until some of O2 Technology's customers prove the company's technology in live applications, though, said Ovum's Jeffcoate, the biggest advantage O2 offers over other OODBMS vendors in Europe and the United States is philosophical. "They are slightly purer in terms of object-oriented technology. The purist view, of course, says that they stand more closely by the object manifesto," she explained. "This may give them an advantage in certain sectors of the market where people care about those things--such as scientific and technological users--but the big commerical companies tend not to care about manifestos."

Bancilhon, however, dismissed the idea that differences between O2 and its competitors are only philosophical, pointing to the quality of O2's user interface tools as one example. Approximately one-third of the $20 million invested in O2 was spent in this area. He added that the money was not spent to create another GUI generator. "We have no intention of competing with them."

Instead, the O2 project aimed to take the objective of OODBMS--that of breaking down the barrier between the database and the programming language--one step further. Now, O2's OODBMS automatically generates a picture on the screen from a logical description of the object in the database. As a result, Bancilhon said, the programmer just writes the application and the interface is generated, breaking down the barrier between the programming language and the user interface. Furthermore, he added, if the programmer is not happy with the data representations given, the system provides tools in O2Look to alter them.

GRAPHICAL MODELING POSSIBLE

France's government agent for all mapmaking in the country identified a different reason to choose O2's product over its competitors. At the Institut Geographique Nationale (IGN), based in Saint Mande, just outside Paris, Benoit David, database administrator for the institute's research lab, studied OODBMS technology in general and the O2 product in particular to see if it was appropriate to handle an application at the institute. The project required modeling and storing of complex geographical data for mapmaking.

David said O2's system, even though it still lacks indexing and clustering functions, lets him do easy modeling of graphical and grid information--something that has always been poorly represented by relational databases. The institute, he noted, tried using two different relational DBMSs--Oracle from Oracle Corp., Redwood Shores, Calif., and Empress from Empress Software, based in Toronto, in its mapmaking applications, but with little success.

IGN gave O2's technology the green light at the end of last year, and David and his staff are currently in the second stage of the study, developing a prototype system scheduled to be ready at year-end.

Complex, geographical data is difficult to represent in relational databases, he said, because they require data to be split into several reference tables. OODBMSs such as O2's can include several references for the same data object, which gives it a more conceptual nature. Geometric information must also have a method associated with it, he added. A list of reference points, for example, must have primitives associated with it such as "line" or "draw" or "point inside," so the system will understand what the points represent. An OODBMS is much more amendable to adding such primitives than a relational DBMS, David said.

Mapmaking data tends not only to be complex, but voluminous. "The originality of O2 is oriented toward large-volume databases. There are a lot more products oriented just [toward] smaller volumes," said David, adding that such products are not useful for their applications.

COMMERCIAL ADVANTAGE

O2 Technology's Bancilhon concurred. "Some [OODBMSs] target smaller databases and that's OK for starting out, for the evaluation phase," he said, "but I believe the object-oriented market is suited to real databases--and large ones."

O2's OODBMS may well prove to be suited to large-volume databases, which would be a commercial advantage, maintained Ovum's Jeffcoate, but it remains to be demonstrated in an operational environment. Some of the modules, such as the O2Tools graphical programming environment, have just gone into beta version [scheduled for April], and functions are still being added to others. The indexing and clustering needed at IGN, for example, is expected this month. The main modules--O2Engine; O2C, an object 4GL; the O2Look toolkit; and O2SQL--are in production versions, however, said Bancilhon.

All of the players in the European OODBMS market are either American or French, Jeffcoate noted. "[France] is very active in object technology, certainly more than either the U.K. or Germany," she added.

In total, the European market comprises a handful of small, startup companies, which are "all doing the same thing on the same [Unix] platform," she said. They include O2 and Graphael from France; Ontos, Inc., Burlington, Mass.; Servio Corp., Alameda, Calif.; Object Design; Objectivity, Inc.; and Versant Object Technology, Menlo Park, Calif., from the U.S. Bancilhon said that O2's biggest competitors are Ontos, Servio and Object Design.

No definite figures on the size of the European OODBMS market are available, though Jeffcoate noted it is a good deal smaller than the North American market. Ovum estimated last year's global OODBMS market, which excludes the Far East, at about $6 million, and believes that this figure will grow to more than $600 million in 1996. Today, only about 20% to 25% of U.S. OODBMS vendors' sales are outside the U.S. In 1996, Europe could account for 40% of the total OODBMS market, or about $240 million, according to Ovum's research.

Also competing in this market, said Jeffcoate, will be the big hardware and relational DBMS companies, such as Hewlett-Packard Co., Cupertino, Calif. In October, HP unveiled an OODBMS called OpenODB, which runs on HP Unix workstations. "I expect that some relational database vendors will add object features to their products, allowing users to upgrade slowly," she said. As a result, Jeffcoate said, some of the startups will survive, but it is unlikely that all six or seven will have a long-term future.

Bancilhon does not believe that he will soon be competing directly with Oracle. "These companies are built on a very strong culture of relational technology. They believe in relational technology and that's it. There's not one example of a computer company built on a single technology being able to switch to another," he said.

When object-oriented DBMS technology first emerged, relational DBMS companies dismissed it as an unwise choice, Bancilhon said. Later, they changed their tune to "objects are easy, we can do it." Lately, he added, they have said they are repainting relational technology in objects.

Such an approach will not work, he asserted, because making a real shift to object technology requires rewriting the database engine. "No one, and certainly not Oracle, will rewrite their core technology for 7% of the market," he said, which is the share of the global DBMS market that O2 forecasts for OODBMSs in 1995.

"Trying to directly address the Oracle customer is not what we're doing. We're trying to offer a service that Oracle cannot offer," he explained. Users involved with technical documentation, telecom network administration and geographical data, for instance, come to O2 saying they need an OODBMS, said Bancilhon.

NEW TARGET: U.S. USERS

Directly addressing the needs of customers of other OODBMS vendors in the United States is high on O2's priority list. The company has already set up an office in Boston to serve as a center for its preliminary activities--taling to industrial partners and beta site setups. Further development of its U.S.-based venture, including direct sales, is the goal of a second round of venture capital financing that at presstime was due to arrive in March. Laurent Hyafil, director of international development at O2, said that the company should have a more complete structure established in the U.S. by summer. The company already has $2 million in financing.

Nevertheless, Ovum's Jeffcoate does not think that O2's Atlantic crossing is such a smart move, citing the failure of compatriot Graphael, which has twice tried to negotiate the U.S. market. "O2's chances of succeeding in Europe are pretty good, but I think it will be difficult in the States. If there was nothing there, they would stand a chance, but with four or five established companies, it could be a nonstarter," she warned.

Bancilhon, whose U.S. background includes a doctorate from the University of Michigan and working at MCC in Austin, disagrees. "We're dedicated to delivering customer training, hotline and other technical support," he said. "We'll succeed [in the U.S.] if we manage to bring something specific to help customers solve their problems."

COPYRIGHT 1992 Wiesner Publications, Inc.
COPYRIGHT 2004 Gale Group