Heralding a new generation of database technology, Burlington, Massachusetts-based software developer Ontologic Inc has released what it clains to be the first available multi-user object oriented database for Unix systems, using the C++ language. The product, called Ontos, was demonstrated running on Sun and Apollo workstations at the recent OOPSLA ’89 exhibition, held in New Orleans. As well as the full-blown Ontos multi-user distributed database, Ontologic is also offering Ontos Persistent C++ – a single user stand-alone version – and a set of extendable class libraries, including aggregates such as sets, lists, dictionaries and arrays. By integrating the programming language and the database, Ontos can outperform relational databases by 10- to 1,000-fold for applications with complex data, according to Robert Martin, vice-president of products and marketing at Ontologic. It depends on the application – computer-aided design and manufacturing, computer-aided software engineering and network management are especially well-suited. However, Martin predicted that eventually all the relational database companies would move towards object oriented products. Some have already added BLOBS, binary large objects, which are nice in some senses, but in other senses are simply flat files. We have flat file performance with database structuring. Martin acknowledged that Unisys Corp’s new Semantic database was a more interesting approach. Ontos currently supports development in C++ and SQL, although SQL limits performance by forcing data into a particular representation model. Work on support for other languages, and on OS/2, DEC VAX and DECstation versions is continuing. Software engineering tools developer Index Technology said it would integrate Ontos into future generations of its tools, and Neuron Data Inc has developed a bridge between Ontos and its Nexpert Object expert system shell. Ontos is priced at $15,000, and will ship by the end of the year.
