Intersystems Corp. has released Caché 5.0, the latest version of its multidimensional database, with a set of features designed for high-performance Web applications.
By expanding support for Java, adding Web services protocols, and boosting performance, the Cambridge, Mass., software vendor hopes to persuade network executives that Caché, rather than traditional relational databases, is the best choice for high-volume e-business Web sites.
Relational databases store data in two-dimensional tables. Often complex queries have to merge, or join, data from numerous tables, bogging down the software. By contrast, databases like Caché store data in multidimensional "cubes," a technique that lets you bundle many data elements together and then very quickly analyze them to discover hidden relationships.
Caché stores data in a multidimensional program, and developers can write traditional SQL queries or use object-oriented programming tools to access it.
Among the additions in Caché 5 are:
-- Transactional bitmap indexing. In effect, you can run very complex queries on transaction data being stored and updated in Caché without bogging down the speed of the transactions.
-- Integrated XML. With a built-in XML parser, the database can create XML documents from its contents, as well as translate the contents of incoming XML documents into Caché structures.
-- Web services. Using the XML parser, Caché can turn any stored procedure, query or database object into a Web service, which can be accessed via Simple Object Application Protocol.
-- Expanded Java support. Caché now automatically creates the code needed to turn a new Java class into an Enterprise Java Bean, which is a component that can be distributed to run on, say, an application server.
-- Enterprise Caché Protocol. The vendor reworked its distributed caching code so it now runs on Java application servers instead of just Caché database servers. The benefit is that data can be stored in memory on these servers for faster Web access and higher traffic volumes.
Intersystems pits itself against traditional database vendors like Oracle. Caché's architecture makes for much faster performance, with much larger amounts of data, compared to relational products, says Paul Grabscheid, vice president for strategic systems at Caché.
The new release stays at the same price, which starts at US$1,000 for five users. In big, multi-server rollouts, says Grabscheid, the cost is about $1,000 per concurrent user. The software is available on most Windows and Unix operating systems, as well as Linux.