|Traditional data warehouses vs. virtual data warehouses
provides a central repository for information
requires the development of metadata definitions
provides data cleansing facilities
uses middleware as data hubs, allowing access to the corporate data stored in heterogeneous data sources
uses middleware to build direct connections among disparate applications
relies on the creation of an independent metadata definition of the corporate data
accesses only data in its raw form unless data access middleware is used to populate OLAP databases directly
One of the key factors governing a company's business success moving into the 21st century is its ability to continue disseminating useful information to decision makers throughout the organization. Today, most large companies already have in place e-mail and collaborative computing systems, such as IBM Corp.'s Lotus Notes, which distribute all kinds of messages and documents throughout the enterprise. In fact, this technology is so widely used that many of us are overwhelmed with information, most of which does little to help us make decisions.
To achieve success in this new millennium, companies must begin to provide true business intelligence (BI). According to market-research firm the Gartner Group Inc. in Stamford, Conn., BI comprises systems that enable a business professional to access the information that describes the enterprise, to analyze it, to gain insight into its workings, and to take action based on its findings, via integration with other office functions. BI systems also present this information in an easy-to-digest form that facilitates effective decision making.
Recently, there has been movement toward virtual data warehouses, which has implications for both information dissemination and improved decision making. Virtual data warehouses allow users to distill the most important pieces of data from disparate legacy applications, without the time, expense, and risk to data required by traditional data warehousing.
Now, as more companies are using Web architectures as the backbone for their enterprise networks, they are moving back toward developing their own information-presentation applications. Many organizations are building or at least considering enterprise portals. An enterprise portal is a single information gateway, typically browser-based, that can be used to navigate and examine both internal and external data, via the Web, and that can have information pushed to it on a regular basis.
While the applications that fit into the portal model are still being explored, it is very likely that a BI application will work very well with this technology. For example, a single enterprise portal could be used to push key business performance information to a firm's management, while at the same time pushing the latest share price of key competitors, along with any related published information from Web sites around the world.
Most of the major BI vendors have enabled their products to be deployed and used over a Web-based architecture. Indeed, they provide a quick solution for deploying Web-based BI applications that are easy to implement. The Web-based development environments and growing strength of eXtensible Markup Language (XML)--the standard Web-based data-manipulation language--make the development and deployment of custom-built applications feasible for most organizations. Many companies are finding that a continuous program of rapidly deploying specific business information monitoring applications can provide key data to allow the company to respond to competitive pressures or market needs. But first you have to get the data
Whether you are focusing on custom-built, Web-based business information systems or packaged online analytic processing (OLAP) tools, the fundamental requirement is to have access to the corporate data to populate these applications. A data warehouse provides an effective method for extending legacy assets while leveraging today's information distribution technology.
Most enterprises employ a wide array of heterogeneous systems, frequently using various databases and file systems. More often than not, in order to provide meaningful business intelligence, data must be mined from more than one of these application data sources. Also, the core business transaction information and the business history is often maintained and stored by legacy applications using legacy data structures.
Because data residing with multiple sources must often be combined for effective decision making, many companies have implemented data warehouses. This creates essentially a hub-and-spoke model whereby each enterprise application sends information to the warehouse, where it can then be accessed throughout the enterprise.