InformationWeek
Defining The Business Value Of Technology
January 5, 1998
Best From Data
Online analytical processing and data mining offer top-down and bottom-up valueBy John Foley with Joy D. Russell
OLAP and data mining support different functions, but both are being more closely integrated with underlying database-management systems. OLAP servers organize data into multidimensional hierarchies, or "cubes," for high-speed data analysis. Data-mining a
lgorithms scan databases to uncover relationships or patterns. The Meta Group, a research firm in Stamford, Conn., says OLAP and data mining are complementary, with OLAP providing top-down data analysis, and data mining offering bottom-up discovery.
The OLAP market grew approximately 40% in 1997, to $1.4 billion, says Richard Creeth, a consultant in Norwalk, Conn., and co-author of The OLAP Report, an online research report (www.olapreport.com). Some market leaders, Creeth notes, are growing at nearly twice the market average.
This year, IBM, Microsoft, and Sybase are expected to introduce OLAP options for their relational databases, joining Oracle and Informix, which already sell OLAP server software.
IBM is teaming with Arbor Software in Sunnyvale, Calif., one of the leading providers of OLAP products, to develop an OLAP layer for its DB2 Universal Database, a relational database-management system. IBM's DB2 OLAP Server, based on Arbor's Essbase OLAP server, is due this quarter. The software will mak
e it possible to run OLAP queries against data stored in DB2, an approach called relational OLAP. By contrast, some OLAP servers require that data be loaded into a proprietary multidimensional database prior to analysis.
Also this year, Microsoft is expected to release an OLAP server, code-named Plato, that works with the company's SQL Server database. Plato is designed to work with a major upgrade to SQL Server, called Sphinx, that is due this year. Analysts say the combination of Plato and Sphinx, which both run on Windows NT, will make more companies look closely at OLAP.
OLAP For Spreadsheets
Among the startups, WhiteLight Systems Inc., in Palo Alto, Calif., introduced its ROLAP server software in December. Sybase, which holds a minority stake in WhiteLight, plans to offer the product as an add-on to its own Adaptive Enterprise Server database. MineShare Inc., in Santa Monica, Calif., plans to release a reporting module, MineShare Publisher, for its own ROLAP suite this month. And Influence Software Inc. in Sunnyvale, Calif., has racked up its first $1 million in sales for its OLAP tool, Aperio, which it introduced last year.
A major effort among OLAP vendors during the year will be to integrate their products with database and applications software. Another focus will be to deliver products that are easier to deploy and manage on the server, and easier to use on the desktop. "The marketplace is not happy with what vendors have delivered," says Kirk Cruikshank, Arbor's senior VP of marketing. "These systems are too ha
rd to deploy."
Arbor's Essbase 5, in beta now and due this year, will make it possible to deploy OLAP applications across multiple systems and WANs. Robert Mondavi Winery in Oakville, Calif., is using Essbase and Essbase Web Gateway for sales, marketing, and financial analysis covering 150 wines. Mondavi deployed sales and marketing applications on Essbase in 1997 and plans to add financial applications this year.
Data-mining vendors will be busy, too. Large computer and software companies will vie-and partner-with smaller ones that specialize in this booming market. According to Meta Group, data mining will grow from $3.3 billion in 1996 to $8.4 billion in 2000. In a survey of large companies, Meta Group found 50% say data mining is critical to their goals over the next two years.
Among developments to look for this year are data-mining products that work in parallel computing environments, dividing the processing job for faster performance, says Aaron Zornes, an analyst with Meta Group. Zornes expects
database vendors to begin adding data-mining functions directly into their database-management systems: "The databases themselves are going to start to take on some of this functionality."
IBM, which introduced a variety of data mining products and services in 1997, will continue the push. Ben Barnes, general manager of IBM's global business intelligence solutions unit, says IBM will commercialize "text-mining" software currently offered on a custom basis. The software, which uses linguistic principles to scan and organize text, has a range of applications, including sorting through call-center records to find trends.
SAS Institute in Cary, N.C., plans to deliver a data mining product, Enterprise Miner, in the first quarter. Last November, SAS announced it will integrate Enterprise Miner with marketing-campaign-management software from Exchange Applications in Boston. The companies say this should eliminate some of the grunt work in using data mining with an applications suite. Fleet Bank, also in Bo
ston, plans to include Enterprise Miner and Exchange Applications' ValEX suite in a data-warehouse environment scheduled to begin in the first quarter of this year.
NCR Corp. in Dayton, Ohio, will integrate SAS Institute's products this year into NCR's Teradata/Scalable Data Warehouse solutions, particularly for industry-specific data-mining solutions. NCR's Neighbor- hood Retailing solutions let retailers analyze data and buying patterns as well as forecast customer behavior. NCR's Relationship Management System provides analytical models and applications to help financial institutions analyze current and potential customers.
Middleware With Data Mining
Another trend: Data-mining tools will become more widely available. At IBM, Bhand
ari helped develop Advanced Scout, a data-mining tool used by some National Basketball Association coaches. Similarly, Bhandari says, easy-to-use products will let many kinds of professionals run data-mining applications against cor- porate data warehouses.
s corporate databases grow, two technologies promise to help companies sort through their terabytes of data: online analytical processing and data mining. Both are sure to be hot sellers this year as companies look for strategic value from their databases.
Microsoft is expected to add OLAP functionality to its Excel spreadsheet and Visual Basic programming toolset. And last fall, Microsoft released a test version of a software specification, OLE DB for OLAP, that will provide some level of interoperability among different companies' OLAP products. More than a dozen companies have said they intend to support it. "OLE DB for OLAP came out of the gate with
wider support than any other standard we've had," says consultant Creeth.
Data-mining algorithms will be included in middleware products that support a range of applications, including E-commerce, knowledge management, and computer security, says Inderpal Bhandari, founder of Virtual Gold Inc., a data-mining startup in Scarsdale, N.Y.
Featured Microsite
