Drill Down: The Business Intelligence Technology Shift

By some estimates, companies will invest as much as $70 billion in business intelligence technology by the year 2000. Researchers estimate that the average cost for a typical data-warehousing project in a Global 1000 company runs about $3 million. And at the bottom line, many projects will fail to deliver their promised benefits. Why? Because not enough people will use the information productively.

 

Business intelligence is the new buzz word used to describe the gathering, management, analysis and distribution of data. The idea is to allow users to extract knowledge from internal and external databases and then apply that intelligence to make strategic decisions, improve marketing, target customers and otherwise improve the bottom line. The applications run from complex data warehouses to single-subject data marts coupled with sophisticated data mining techniques to online analytical processing analysis.

From Data-Centric to User-Centric

But regardless of the underlying technology supporting the business intelligence, the key to realizing the expected return on investment in a data warehousing/business intelligence project rests on the ability of the targeted users to easily analyze and effectively access the available data. Although the early pioneers believed that if they built a data warehouse, the users would come, that sadly has not proven to be the case. Yes, users will come, but only if the path is pretty much free of obstacles.

Consequently, within some quarters of the business intelligence community there has been a shift from a data-centric to user-centric perspective. From a data-centric vantage point of view, the emphasis is on the effective, appropriate and logical organization of data. The user-centric approach demands that access to data be fast, easy and relevant to the user.

Fat vs. Thin Clients

Interestingly enough, the user-centric approach focuses attention on the client software side of the equation. In most client-server applications, the real attention is to the server software. The client is the interface that allows the user to tell the server what to do.

Over the past several years, the widespread deployment of the World Wide Web has raised a number of controversies concerning client software. The first is the debate over fat clients versus thin clients. A fat client is one that requires the computing resources of a personal computer. A thin client consumes fewer resources. The debate was sparked when Sun first started aggressively marketing Java. The thought was that Java applets running on stripped down computers could deliver enough muscle to server as the client in many client-server applications. The argument was that thin clients would be less expensive and easier to deploy. Sun Microsystems - certainly not a disinterested party - claims that its internal experiences indicate that running Java-based clients in an enterprise environment costs $2,500 a year per seat compared to $10,000 to $15,000 per year per seat for "fat" (read Wintel) clients.

Despite the emotionally laden terms of the debate - at least in America "fat" is automatically bad and "thin" is automatically good - the issue has not been put to rest. Thin clients have not replaced fat clients in many mainstream applications.

The World Wide Web also raised a second issue concerning client software - it’s pricing. Regardless of the Justice Department’s attack on Microsoft, the tradition in the Web is for the client (the browser in this case) to be free. This not just true for Microsoft’s Internet Explorer. Every company offering a specialized media type on the Web such as Real Networks audio and video information and Adobes Shockwave offer browsers for free.

Web Access

Those two issues have come to bear in client software in the business intelligence and data-mining arena as well. In place of dedicated clients to access servers, many companies wish to provide Web access to data warehouse information. Since Web browsers have become virtually universal, using the Web infrastructure to open data warehouses to end users seems natural. It eliminates the installation and administration of data analysis tools on every client machine. It fits the requirements of mobile professionals. And it cuts down on the need to train and support users. For those compelling reasons, analysts project that 60 percent of data warehouse projects over the next year will incorporate Web-enabled technology.

But despite the promise of Web access, its implementation is not problem-free. A Web-based solution requires a browser to send instructions to a Web server and then to a Web gateway that translates and passes the query to a data warehouse server. The process is reversed on the way back. When the Web server sends the request page to the browser, the connection to the user is then closed.

As there is no ongoing connection, the interaction between Web server and browser is called "stateless." Stateless interactions work well for static information, in dynamic applications such as data mining, repetitive queries increase network traffic and poorly utilize bandwidth.

The Web presents other problems as well in terms of security, data entry and reporting. To address those problems, companies, such as Hyperion Software, are creating Java applets to address those needs. Applets sent by the server can manipulate data locally, performing tasks the browser on its own cannot.

Free OLAP Client

If to utilize Web access to data warehousing requires deploying Java applets to overcome the shortcomings of browser technology, at least one vendor of a full-featured OLAP client has responded to the pricing issue. This quarter, Seagate Software released Seagate Worksheet, an OLAP analysis client designed to be used with a wide range of multi-dimensional OLAP databases including Microsoft DSS (Plato), Arbor Essbase, Seagate Holos, IBM DB2 OLAP Server or any OLE database provider. The cost? Nothing at all; it is free.

According to Seagate Software officials, the market is awash with undifferentiated OLAP analysis client tools. Consequently it makes sense to offer it for free. As companies build their business intelligence infrastructures, hopefully they will turn to Seagate Software’s other offerings.

Conclusion

One of the lessons of the Internet is that access is everything. To get people to use your information, you must give away the browser. The concept is slowly moving into the data-warehousing environment. As the technology matures, it should make more information more useful to more people.

 

 

ABOUT THE AUTHOR:

Dr. Elliot King is an Assistant Professor of Communications and Director of the New Media Center at Loyola College in Maryland. His research interests are in distributed information systems, new communications technology and the diffusion of innovation. He has written five books and hundreds of magazine and journal articles about the use of new information technology. He can be reached at (410) 356-3943, by fax at (410) 356-5217, or by e-mail at eking@loyolanet.campus.mci.net.