Tandem aims to carve itself a niche in data mining and to rally industry heavyweights behind its technology, announced on Tuesday.
As part of its Object Relational Data Mining (ORDM) initiative, the company is redesigning the database at the heart of its Serverware enterprise software to be optimised for mining. It is also building an extension to SQL for mining, which it is lobbying to have incorporated in the upcoming SQL 3 standard.
It already has support from Informix and probably from Microsoft in this attempt, but Roel Pieper, chief executive of Tandem, said he expects Oracle to "go its own way". This highlights the battle lines that are being drawn in the data analysis market, with Oracle and Microsoft pursuing conflicting strategies in online analytical processing and mining. A Tandem source said the company is confident of Microsoft's backing because the two companies cooperate heavily already in areas such as clustering.
The ORDM blueprint allows the entire data warehouse to be available to the mining tools and sets up a standard SQL interface between client mining tools and both the object and relational database engines. The data mining algorithms are then embedded into this database architecture.
"Initially the use of SQL extensions will greatly enhance the way traditional alphanumeric data types are mined," commented development director Abhay Mehta. "As technology evolves, this architecture will enable fast, efficient mining of more complex data types such as image, voice, video and other multimedia objects. In the second half of 1997 Serverware will be the first database to combine all the elements."
The SQL extensions will be written as Data Blades, the software modules that work with Informix' Universal Server database. They will enable data mining tools to work with any supporting database without rewriting.
The first data mining function to be embedded in Serverware will be available in the third quarter and will establish cross tables to simplify analysis patterns. Future additions will include mining algorithms such as neural networks.
Tandem claims the initiative will appeal to its core customer base of large corporates, especially banks, because decision support and operational data will be centralised on one server. As well as the database, Serverware also incorporates a transaction processing monitor and messaging middleware. "This combination is the aim of the whole object-relational movement but we are delivering real solutions for real companies," said Tandem.
Following its recent alliance with Microsoft, Serverware will work with NT as well as Tandem's own systems.
Several data mining and decision support vendors announced that they would incorporate Tandem's SQL extensions in their tools. These include Angoss Software and Neovista Software, among others.
Data mining tools, which analyse trends within data warehouses using techniques such as pattern recognition or neural networks, take analysis techniques such as Olap a step further. They are predicted to grow as a market from the current value of $1.5 billion to $6 billion in 2000, according to the Meta Group.
Campaigners want US authorities to break-up Instagram, WhatsApp and Messenger into separate companies
The perception of the industry as "a white man in a hard hat" is limiting new applicants, says Hayaatun Sillem
Almost two years late - and just as AMD is readying 7nm Zen 2 for early 2019
Eye-wateringly expensive smart speakers take just six per cent market share, claims Strategy Analytics