MetaCube 3.0 is the first release of the company's online analytical processing server tuned to the Informix database since the product was obtained through the acquisition of Stanford Technology Group last fall.
OLAP servers are specialized databases that analyze data extracted from relational databases. The servers help users with answers to frequently asked business questions and perform data analyses, such as calculating sales figures for a specific region or forecasting consumer buying trends.
Informix also a relational database server for general purpose business applications called Online Dynamic Server. The company plans to integrate function-specific servers, such as MetaCube, into the database beginning next year to give users a single database for performing both data retrieval and analysis.
The first version of that redesigned database, called Universal Server, is expected to ship on Unix operating systems by year's end, and on Windows NT early next year. MetaCube will be integrated into Universal Server by mid-1997, said the company.
Release 3.0 of MetaCube is the first step toward an integrated database server, said Bruce Golden, general manager of Informix's data warehousing development unit. "We're moving the OLAP technology into the database itself. MetaCube 3.0 is an intermediary step toward building that," said Golden.
MetaCube 3.0 takes advantage of how Online Dynamic Server processes data to improve performance, according to the company. Previous versions were more generalized and designed to work with relational databases from Informix, Oracle, and Sybase.
Informix has also added a new capability for customizing the way MetaCube processes data. New components, called Snap-Ins, can be plugged into the MetaCube 3.0 engine to perform specific functions, such as calculating average gross margins or other calculations. Informix is recruiting third-party software developers to build the Snap-Ins, and a Snap-In software development kit will be included with MetaCube 3.0.
Another new feature is data sampling. Instead of performing calculations against many megabytes of data, MetaCube 3.0 can extrapolate an average from a small sample of data. "It can take a sample like one out 100 or 1000 records and predict a result or draw a trend, as is done with predictions in presidential elections," said Glen Kelman, a product engineer at Informix.
MetaCube 3.0, now in beta testing, is slated to ship later this year. It is priced from $500 to $1,000 per user.