By Mitch Kramer, Green Hill Analysis icon

By Mitch Kramer, Green Hill Analysis

Green Hill Analysis

A Comparison of Business Intelligence Strategies and Platforms

Comparing Microsoft, Oracle, IBM, and Hyperion

By Mitch Kramer, Green Hill Analysis

November 2002

Prepared for Microsoft Corporation


This report evaluates and compares the business intelligence platform strategies and business intelligence platform components of Microsoft Corporation, Hyperion Solutions, Inc., IBM Corporation, and Oracle Corporation. The evaluations and comparisons are made through the analysis of the components of business intelligence platforms: data warehousing databases, OLAP, data mining, interfaces, and build and manage capabilities. All four vendors have characteristic strengths and limitations. Relative to its competitors, Microsoft has a few disadvantages and several significant advantages.

© 2002 Green Hill Analysis. All rights reserved.

The information contained in this document represents the current view of Green Hill Analysis on the issues discussed as of the date of publication. Because technology and market conditions are constantly changing, Green Hill Analysis cannot guarantee the accuracy of any information presented after the date of publication.

Microsoft, Windows, SQL Server, .NET, Visual Studio, and Commerce Server are either trademarks or registered trademarks of Microsoft Corporation in the United States and/or other countries

Other product or company names mentioned herein may be the trademarks of their respective owners.

Green Hill Analysis, 21 Metacomet Way, Sudbury, MA 01776 USA


Table of Contents

Green Hill Analysis i

By Mitch Kramer, Green Hill Analysis i

November 2002 i

Prepared for Microsoft Corporation i

What Is Business Intelligence?

Business intelligence has become a critical element of information technology. It’s an old term with general or even ambiguous meaning. It has been used synonymously with decision support, analysis, and data warehousing, but today business intelligence has a more specific definition and a better understood application. Taken literally, business intelligence is just that—intelligence or understanding of your business. You get that understanding by analyzing your business operations.

That analysis is accomplished by collecting the information that represents your marketing, sales, and service activities, the behavior of your customers in responding to these activities, and the behavior of your internal systems and your suppliers’ systems in responding to your customers’ behavior. Once you have collected this information, and its collection is a continuous process, not a one-time event, you organize and store it in a manner to facilitate its access, processing, and presentation through a broad range of techniques including, reporting, query and analysis, OLAP, and data mining. Finally, you use the results of applying these techniques to improve your business operations and start the analysis cycle all over again.

This business intelligence process can deliver significant, bottom-line results. Implementing its technologies and applying its process can help make your business more effective and more efficient, increasing revenue, decreasing costs, and improving your relationships with customers and suppliers.

The Evolution of Business Intelligence

Business intelligence technologies and business intelligence usage have also become better understood. They have been more efficiently implemented and more effectively applied, too. It wasn’t so long ago that business intelligence was implemented by a loose collection of technologies, deployed only by those companies that seem always to install the latest technologies, applied in ad hoc ways, and used by only a few individuals who were interested enough to develop the skills necessary to use and apply these technologies. We saw pockets or silos of business intelligence technologies and their applications. Benefits achieved were narrow, but potential benefits appeared quite broad.

In fact, in 2000 we wrote a predecessor to this report that focused on OLAP, one in that loose collection of business intelligence technologies. We evaluated and compared the OLAP capabilities of its leading suppliers and examined other business intelligence technologies that they offered.

Today, business intelligence technologies are more tightly integrated and more easily and more widely deployed and used. Business intelligence applications reach to the edges of corporations and beyond corporate boundaries to customers and suppliers.

The current economy has been major driver for these improvements in business intelligence. We are operating in an economic climate that demands more careful justification of technology investments and accelerated returns on them. Companies want to use technology tactically to make their operations more effective and more efficient. Business intelligence can be the catalyst for that efficiency and effectiveness. And, business intelligence has become so much easier to justify and demonstrate accelerated returns.

Business Intelligence Platforms

In order to deliver business intelligence to the widest audience and to maximize the benefits that it can deliver its technologies must be organized. They must be deployed within an infrastructure with the capabilities to implement the business intelligence process that we described above and to support the range of applications best suited to every user of every type. We call that infrastructure a business intelligence platform.

Business Intelligence Platform Requirements

Business intelligence platforms should include the following technologies. Each technology should implement the capabilities described below.

  • Data Warehouse Databases. A business intelligence platform should support both relational and multidimensional data warehousing databases. In addition, storage models should support the distribution of data across both and data models should support transparent or near-transparent access to data, wherever it’s stored.

  • OLAP. OLAP is a critical business intelligence platform component. It is the most widely used approach to analysis. Business intelligence platforms must provide OLAP support within their databases, OLAP functionality, interfaces to OLAP functionality, and OLAP build and manage capabilities.

  • Data Mining. Data mining has reached the mainstream. It is a critical business intelligence platform capability. Platforms should include data mining functionality that offers a range of algorithms that can operate on data warehouse data.

  • Interfaces. Business intelligence platforms should provide open interfaces to data warehouse databases, OLAP, and data mining. Where appropriate, interfaces should comply with standards. Open, standards-based interfaces make it easier both to buy and to build applications that use the facilities of a business intelligence platform.

  • Build and Manage Capabilities. Business intelligence platforms should provide the capabilities to build and manage data warehouses in their data warehouse databases. Build capabilities should include the implementation of data warehouse models, the extraction, movement, transformation, and cleansing of data from operational sources, and the initial loading and incremental updating of data warehouses according to their models. A wide range of data sources should be supported including databases, files, and the data of popular packaged software. Transformation capabilities should be powerful and flexible. Predefined transformations should be packaged. They should be extensible through programming languages. Manage capabilities should cover all platform resources—users, data, and processes. Strong and flexible prepackaged capabilities are essential. Good use should be made of visual tools

The business intelligence platform should provide good integration across these technologies. It should be a coherent platform, not a set of diverse and heterogeneous technologies. For example, a single toolset should provide build and manage capabilities across both relational and multidimensional data warehouses.

The Role of Partnerships

Partnerships are essential for the viability of all software suppliers. No software supplier can expect to address all requirements through the resources of its own R&D organization and, at the same time, remain profitable and competitive. Through partnerships, software suppliers can get to market faster and can respond more quickly to market changes. Through partnerships, software suppliers can specialize their R&D resources, making them more productive and more agile.

For business intelligence platforms, the approach to partnerships differentiates suppliers. Partnerships indicate the level of commitment to the business intelligence platform and the level of control over the technologies that comprise these platforms. We believe that any business intelligence platform supplier should own all its platform’s technologies as well as its build and manage capabilities. Suppliers can more appropriately use partnerships for the end-user tools to access platform resources and to build custom business intelligence applications. Suppliers should also use partnerships for the packaged business intelligence applications that use the resources of their platforms.

The Leading Suppliers of Business Intelligence Platforms

In our 2000 analysis and comparison of OLAP, the leading OLAP suppliers were (alphabetically) Hyperion Solutions, IBM, Microsoft, and Oracle. In this analysis and comparison of business intelligence platforms, the leading suppliers are the same. They’re the companies that the market relies on for business intelligence infrastructure.

What has changed over the past two years are their leadership positions. For example, in 2000, Microsoft was the comer. Relatively new to both OLAP and business intelligence, the firm’s offerings were just beginning to get traction. Microsoft made great technology decisions and investments in the late 1990s. Today, Microsoft is quantitatively the OLAP leader and its business intelligence platform is the equal of any of the leaders. Also, the pricing and packaging advantages that Microsoft demonstrated with OLAP in 2000 today are demonstrated for its entire business intelligence platform. As a result, this platform delivers value that is not approached by the platforms of the other leading suppliers.

In 2000, Oracle offered well-proven but acquired OLAP capabilities. But, these OLAP capabilities were implemented separately and independently of Oracle’s relational capabilities. Over the past year or two, Oracle has moved to a more integrated business intelligence platform by delivering both OLAP and relational capabilities in its relational database. As a result, today, Oracle offers a more technologically consistent business intelligence platform, but its OLAP implementation has not yet been widely adopted by tools and application suppliers and, therefore, hasn’t yet achieved significant market share.

Hyperion was the OLAP leader in 2000. From a business intelligence platform perspective, this firm offered an OLAP database, the tools to support it, and the interfaces to support OLAP applications. Today, while a new version of its Essbase OLAP server was introduced in April 2002, Hyperion’s business intelligence offerings are essentially the same. The company has not evolved with changing platform requirements. It now shares OLAP leadership with Microsoft, but the narrowness of its offerings will prevent business intelligence platform leadership.

IBM has improved its position since 2000, integrating data mining more tightly into its business intelligence offerings. But, IBM still relies on an OEM partnership Hyperion for OLAP to complete its business intelligence platform. This partnership prevents IBM from controlling its platform and demonstrates less than a complete commitment to it.

This Report

This report has two sections. In the first section, we will describe and analyze the business intelligence platform strategies of the four leading suppliers, discussing their strategies, partnering approaches, and packaging and pricing as well as briefly describing the products and technologies that make up their platforms. In the second section, we’ll evaluate and compare how the four suppliers address each of the business intelligence platform requirements, identifying key advantages and disadvantages.

Comparing Business Intelligence Platform Strategies

Microsoft’s Business Intelligence Platform Strategy

Microsoft’s business intelligence platform strategy is rooted in its database offering: SQL Server 2000. SQL Server 2000 is the anchor storage and query technology behind.NET servers. And, one of the key applications for SQL Server is business intelligence. The business intelligence platform strategy for Microsoft leverages SQL Server-based technologies and products through these elements:

  • Deliver a comprehensive business intelligence platform with advanced data warehousing techniques, great analytic functionality, and excellent performance and scalability across all platform components

  • Through Microsoft’s business intelligence platform:

  • Push business intelligence to the edges of the enterprise

  • Make business intelligence more pervasive within the corporation

  • Make business intelligence more reachable for more users and more types of users

Microsoft implements this platform strategy through the application of a familiar and well-proven Microsoft product marketing formula. That formula has these key elements:

  • Fast implementation

  • Ease of learning and ease of use

  • Low cost and high value

  • Fast return on investment (ROI)

This is the formula that Microsoft has repeatedly demonstrated and consistently proven. The company has used it successfully for its Windows platform (now .NET platform), its SQL Server database as used for OLTP applications, its e-commerce platform, Commerce Server, and its application development suite, Visual Studio. While all of us have been conditioned to be skeptical (even doubtful) about phrases like “fast implementation” and “ease of use,” Microsoft has always delivered on them. And, this is not just a vision for small organizations with small budgets. Microsoft delivers data warehousing value to companies of all sizes.

Netting it out, Microsoft’s business intelligence platform strategy enables companies of any size, of any level of business intelligence skill and experience, of any IT budget to deploy business intelligence throughout and to deliver and achieve the benefits of business intelligence—improved effectiveness, greater efficiency, and higher quality throughout all their business processes.


Microsoft’s business intelligence platform is built on Microsoft technology. Microsoft controls all aspects of it design, development, product marketing, and support. The firm feels that this platform is core to its database strategy and an integral component of .NET. Control of the entire platform from planning, design, development, and product marketing perspectives is essential in order to provide consistency, integration, and timely technology delivery for both Microsoft’s customers and its partners. However, partnerships are critical to Microsoft and to its business intelligence platform. The firm currently uses partnerships to create a large base of specialized business intelligence tools and applications that supports its platform. These partnerships simplify and accelerate adoption of the platform and make the platform’s resources more easily accessible. This is an ideal partnering approach.

Contrast Microsoft’s approach with IBM’s. Where Microsoft owns its business intelligence platform, IBM OEMs its platform’s OLAP technology from Hyperion, and, as a result, IBM doesn’t control its business intelligence platform. Or, contrast Microsoft’s approach with Oracle’s. After maintaining separate and independent relational and OLAP technologies and product offerings since its 1995 acquisition of Information Resources, Inc. and its Express technology, in Oracle9i, Oracle has implemented OLAP within its RDBMS and has de-emphasized Express. This approach gives Oracle the advantage of control over its business intelligence platform, but it’s an approach that has the current disadvantages of immaturity and limited market adoption.

Microsoft’s Business Intelligence Platform

Microsoft’s business intelligence platform is built on SQL Server. SQL Server features provide relational and multidimensional data warehousing, OLAP, data mining, and build and manage capabilities for relational and multidimensional data warehouses. SQL Server also provides an array of application interfaces all built on the flexible and extensible object-oriented COM component model. These interfaces provide the access to all business intelligence resources with the flexibility and control to address any application requirement. In fact, almost all of our requirements for business intelligence platforms are addressed completely by features of SQL Server.

Packaging and Pricing

Packaging and pricing distance Microsoft’s business intelligence platform from the platforms of Oracle, IBM, and Hyperion. For the processor-based license fee of $19,999 per processor for SQL Server Enterprise Edition, you get the entire business intelligence platform. OLAP, data mining, and build and manage capabilities are included as database features.

Oracle charges $40,000 per processor just for the Enterprise Edition of its relational database. OLAP, data mining, and build and manage capabilities are all separately priced and packaged features of this Enterprise Edition of the firm’s database. IBM charges $25,000 per processor just for the Enterprise Server Edition of its relational database with included but basic build and manage capabilities. OLAP, data mining, and advanced build and manage are all extra. Hyperion charges $28,000 per processor just for OLAP with packaged build and manage capabilities.

Oracle’s Business Intelligence Platform Strategy

Oracle is a software company with two major lines of business: databases and applications. The current flagship offering of the firm’s database business is Oracle9i. This is an object/relational database management system designed and positioned to support all types of Internet-based applications. Oracle9i integrates what Oracle terms a “complete and integrated infrastructure for building business intelligence applications.” So, Oracle’s business intelligence platform strategy to provide a comprehensive business intelligence platform built on and integrated within its flagship database system.


Partnerships do not play a major role within Oracle’s business intelligence platform. The “complete and integrated infrastructure” means that every platform component is provided by Oracle and is based on Oracle9i. Oracle’s R&D organization controls the design, development, and support of the entire platform. Like Microsoft, Oracle uses partnerships for business platform tools and applications, although the firm currently competes with these partners with its own business intelligence tools and applications. Because Oracle’s business intelligence platform does not include partner technology, Oracle has the important advantage of control over the platform’s components, technology, and integration.

Oracle’s Business Intelligence Platform

Oracle9i is the foundation of Oracle’s business intelligence platform. OLAP functionality is provided by Oracle9i OLAP and data mining functionality is provided by Oracle9i Data Mining. Both are features of Oracle9i Enterprise Edition. Build and Manage functionality is provided by two toolsets. The first, Oracle Enterprise Manager, is the main management framework and DBA toolset as well as the toolset for OLAP build and manage. The second is Oracle9i Warehouse Builder. This toolset, a component of Oracle Internet Developer Suite, provides capabilities for managing relational data warehousing resources, designing relational data warehouse models, and ETL.

Packaging and Pricing

From a packaging perspective, Oracle offers little bundling. All the components of its business intelligence platform are separately packaged and priced and the build and manage components have separately priced and packaged sub-components. Oracle9i Enterprise Edition is priced at $40,000 per processor. Oracle9i OLAP is priced at $20,000 per processor. Oracle9i Data Mining is priced at $20,000 per processor. And, Oracle Warehouse Builder is priced at $5,000 per named user. Add them up and Oracle’s business intelligence platform is at least five times higher in price than Microsoft’ business intelligence platform.

IBM’s Business Intelligence Platform Strategy

From a corporate perspective, IBM has three businesses: hardware, software, and consulting services. The software business has four components: WebSphere software, DB2 data management software, Lotus (collaboration) software, and Tivoli (system management) software. Business intelligence is one of two IBM-provided solutions of DB2 data management software. (The other solution is e-business.)

IBM defines business intelligence as “warehousing, data mining, and OLAP.” That’s exactly our definition of a business intelligence platform. So, IBM’s business intelligence solution is a business intelligence platform.

IBM’s strategy for business intelligence is to help companies know their customers and to use that knowledge to gain competitive advantages, to maximize revenue, and minimize cost. Business intelligence is implicitly targeted at all of IBM’s markets. The firm makes no explicit distinction in the positioning of business intelligence for the types or sizes of companies or for the types of users within those companies that can use its business intelligence platform. It’s a one size fits all approach.

IBM’s Business Intelligence Platform

IBM’s business intelligence platform is based on its DB2 data management software. The DB2 Universal Database (UDB) provides relational data warehousing capabilities. The database also integrates basic relational data warehousing build and manage capabilities. OLAP functionality and OLAP build and manage functionality are provided by DB2 OLAP Server, a feature of DB2 Enterprise Server Edition that is OEMed from Hyperion and re-branded. Data mining functionality is provided DB2 Intelligent Miner and DB2 OLAP Miner. The integrated relational build and manage functionality of DB2 is enhanced with Warehouse Manager and DB2 OLAP Administrative Services provides OLAP build and manage capabilities.


IBM relies on a set of partners to assemble it business intelligence platform. The most critical partnership is with Hyperion Solutions. Partners also provide ETL and data cleansing capabilities that augment the platform’s build and manage capabilities.

The partnership with Hyperion, formed in 1999, gave IBM OLAP capabilities instantly, allowing IBM to compete in an important market where it had no previous presence and where it had made no investment in R&D. For the long term, however, this partnership is a disadvantage to IBM’s business intelligence platform and to its business intelligence customers and partners. Why? OLAP is more an add-on than an integral component of IBM’s business intelligence platform. IBM has no direct control over OLAP technology, its development, and its integration within its business intelligence platform. Product development schedules cannot be synchronized. IBM build and manage technologies cannot easily be extended to address OLAP as well as DB2 data warehouses. Given the importance of OLAP, IBM should either acquire Hyperion or develop its own OLAP. Until then, its business intelligence platform will always be at a disadvantage.

Packaging and Pricing

With three exceptions, every component of IBM’s business intelligence platform is separately packaged and priced. While individual components offer good value, the platform, as a whole can be quite costly. DB2 UDB is priced at $25,000 per processor for the Enterprise Server Edition V8.1. It includes the basic build and manage functionality of Data Warehouse Center. DB2 OLAP Server is priced at $28,000 per server with an additional $1,500 fee per named user. It includes the build and manage capabilities of DB2 Administrative Services and the data mining functionality of DB2 OLAP Miner. Intelligent Miner has three components. When you buy all three components, you’ll pay a total price of $75,000 per processor. And, for advanced relational build and manage capabilities, DB2 Warehouse Manager is priced at $10,600 per processor. That adds up to $138,600 before you add in the prices for external ETL and data cleansing tools and per user fees for DB2 OLAP Server. Compare that with the $19,999 per processor for the entire Microsoft business intelligence platform.

Hyperion’s Business Intelligence Platform Strategy

In mid 2001, Hyperion changed its corporate strategy, shifting its focus from business intelligence software infrastructure and applications to business performance management software solutions. The firm states its objective “is to be the leading global provider of business performance management solutions.” These solutions are designed to automate the business performance management process of strategy setting, modeling, planning, performance monitoring, reporting and analysis. Their objective is to improve your profitability.

The technology platform for Hyperion’s performance management solutions is Essbase, its venerable OLAP Server. Within the new strategy, Hyperion states that Essbase technology will be enhanced in the areas of ease of use, ease of application development, interoperability of business performance management applications, scalability, and tighter integration with relational data sources. Missing from Hyperion’s enhancement strategy for Essbase are areas such as analytic technology and business intelligence platform technologies. Essbase is evolving away from a general purpose OLAP facility and toward a platform for supporting a very specific type of business intelligence application.

Hyperion’s new strategy and objective can play quite well in today’s business intelligence market where companies’ top priorities are to do business more effectively and efficiently. Business performance management is a classic business intelligence application. It requires a comprehensive business intelligence platform as its foundation in order to collect the information that represents business performance, organize that information, analyze it, present it, and use analysis results to improve business performance.

In Essbase, Hyperion provides a critical component of a business intelligence platform, and for business performance management applications as Hyperion defines them, Essbase can be the leading platform component, the first component to be implemented. However, additional platform components not offered by Hyperion are required, most significantly relational databases, data mining tools and analytic applications. As a result, while Hyperion’s strategy can make it a leader in an important class of business intelligence applications, its platform strategy makes it a niche player in business intelligence platforms.

Hyperion’s Business Intelligence Platform

Hyperion does not provide a complete business intelligence platform. Rather its Essbase product can provide the OLAP functionality within a comprehensive business intelligence platform. Essbase also integrates OLAP build and manage functionality.


Hyperion must partner in order to expand from its OLAP niche and address all business intelligence platform requirements. IBM is Hyperion’s most important business intelligence platform partner. Hyperion also has many business intelligence tools and applications partners that leverage the OLAP capabilities of Essbase.

Packaging and Pricing

Hyperion Essbase has a pricing model for with two elements: a per server fee and a per named user fee. Currently, the per server fee is $28,000 per processor and the per named user fee is $1,500. Essbase packaging includes the OLAP server, administrative tools, and build and manage tools.

Essbase installations most commonly use relational data warehouses as the data sources for Essbase cubes. This approach requires the purchase, implementation, and support of a complete data warehousing infrastructure in addition to the purchase, implementation, and support of Essbase and its associated tools in order to do OLAP business analyses. The price of Essbase is roughly equivalent to the price of a relational data warehouse database. So, Essbase installations buy two data warehousing database products. Until Microsoft included OLAP capabilities with its relational database, this has been the case and continues to be the case with all OLAP implementations.

Comparing Business Intelligence Databases

Microsoft, Oracle, and IBM all offer relational databases as the foundation for their business intelligence platforms: Microsoft SQL Server, IBM DB2 Universal Database (UDB), and Oracle9i. The databases all provide excellent relational data warehousing capabilities as well as integrated OLAP data warehousing capabilities, data mining capabilities, and build and manage capabilities.

Overall, it’s beyond the scope of this report to compare these offerings on their relational data warehousing capabilities. Suffice to say, all three are viable relational data warehousing offerings. However, their approach to integration of other business intelligence platform functions and their implementations of those functions are the areas where significant differentiation becomes apparent. We’ll briefly discuss the relational business intelligence capabilities in this report section. We’ll get into the differentiated areas in the sections following.

Note that Hyperion does not provide a business intelligence database. Hyperion offers its Essbase OLAP platform more as a complement to relational business intelligence platforms than as a platform, itself. Most installations implement a relational data warehouse and source multidimensional Essbase structures from it.


SQL Server 2000 is Microsoft’s business intelligence database and the foundation for all of the components in Microsoft’s business intelligence platform. All of the company’s business intelligence platform technologies and products are implemented as SQL Server 2000 no-charge features and are included “in the box” with this relational database—build and manage facilities, OLAP, and data mining.

As a relational database for data warehousing, SQL Server has been improved significantly. It has always had usability advantages over IBM and Oracle for data warehouse implementation, build and management, and performance tuning. Lack of performance and scalability, especially, scalability, has been its limitation historically, but architectural improvements to its database engine, data warehousing features and a big boost from fast SMP hardware have enabled SQL Server to compete with IBM and Oracle across most of the scalability curve. The sweet spot for SQL Server in business intelligence ranges from the low-end the middle and touches the high end in terms of capacity and scalability. When you factor in pricing and packaging, SQL Server delivers business intelligence value that its competition doesn’t approach.


Oracle9i is Oracle’s business intelligence database. 9i is an object/relational database that has long packaged excellent data warehousing features, especially so for top-end applications. The best of these top end features are a wide range of index types, rich join capabilities, and multiple approaches to partitioning.

Oracle9i is also becoming a comprehensive and well-integrated business intelligence platform. In Oracle9i Release 1, which was introduced in April 2001, Oracle added OLAP capabilities within its database as well as enhancing build and manage capabilities to support multidimensional warehouses and marts. In Release 2, which was introduced a year later, OLAP capabilities were improved and data mining capabilities were added.

Oracle9i now has the advantages of comprehensiveness and integration for all aspects of a business intelligence platform. However, the database also has disadvantages as a business intelligence platform. First, both OLAP and data mining capabilities are newly implemented in the database and neither is well proven or widely used. Second, while the capabilities are comprehensive and they’re offered as database features, they’re separately packaged and priced, adding considerably to the initial platform cost.


DB2 Universal Database (UDB) is IBM’s strategic business intelligence database. IBM now offers versions of DB2 UDB for Windows, for the leading Unix platforms, and for its proprietary S/390 mainframe and AS/400 midrange platforms. DB2 is a strong offering that provides good top end support for business intelligence applications, although, the product scales down very well and provides good usability in its administrative tools.

DB2 OLAP Server provides the business intelligence database for OLAP. As we’ve previously mentioned, DB2 OLAP Server is Hyperion’s Essbase, OEMed, integrated, and re-branded by IBM. It’s not tightly integrated within the IBM business intelligence platform. It requires separate build and manage tools and is not supported by IBM’s mainline data mining offering. However, new in its latest release V8.1, multidimensional storage and data models can be allocated across DB2 OLAP Server and DB2 and IBM has added OLAP-based data mining.

Note that in July 2001, IBM acquired the database technology of Informix Software. This technology includes the Informix RDBMS family and the Redbrick data warehouse database, both viable business intelligence databases. To date these offerings and their build and manage capabilities have remained separate from DB2 and excluded from IBM’s business intelligence platform strategy. IBM needs to articulate and implement product and marketing strategies for these database offerings. Customers are confused. They need direction from IBM.

Comparing OLAP

The OLAP functionality of all four leading business intelligence suppliers is quite similar and quite good. We’ll not get into too much detail on OLAP functionality in this report, but there are characteristics of these OLAP implementations that really differentiate them from each other. These characteristics are packaging and pricing, integration into the business intelligence platform, and maturity and market adoption. These characteristics for the four suppliers are:

  • Microsoft has included OLAP functionality within the no-charge SQL Server Analysis feature of its SQL Server database. The feature has been available since 1998.

  • Oracle9i OLAP is a separately packaged and priced database feature. Oracle9i Release 1 OLAP was introduced in April 2001. Oracle9i Release 2 OLAP was introduced in March 2002. Release 1 OLAP did not offer viable capabilities. Release 2 OLAP is not yet widely used or well proven.

  • DB2 OLAP Server is OEMed from Hyperion, re-branded and offered as a separately priced and packaged feature. DB2 OLAP Server was introduced in 1999. Its Essbase technology, introduced many years earlier, is widely used and well proven.

  • Essbase is a comprehensive, dedicated OLAP system. Essbase was introduced in 1992. Essbase technology is widely used and well proven. It’s the most mature OLAP of the four of those discussed here.

Microsoft Corporation

SQL Server Analysis Services OLAP functionality includes the building and management of multidimensional OLAP data models, the implementation loading and updating of those models in easily configurable MOLAP, ROLAP, and HOLAP stores, a large set of predefined data access and analytic functionality, and application interfaces to this functionality.

Quantitative analysis functions are a strength and differentiator of Analysis Services. These functions include statistical processing capabilities and the capability to execute a data mining model. In addition, user-defined functions are supported and Analysis Services provides the documentation and tools for developing them. That’s good power and flexibility.

Two other features that make Microsoft’s OLAP offering particularly flexible are actions and custom roll-ups. Actions are like relational triggers but for multidimensional data. They can extend analyses to incorporate custom functionality or they can even be used to close the loop between analytic applications and operational systems. Custom roll-ups allow the values in a parent dimension values to be calculated from the values of their children individually, one member at a time, rather than by conventional aggregation. Custom roll-ups enable the implementation of business domain-specific analysis.

Microsoft’s OLAP offering is impressive considering that OLAP was a new technology offering in 1998 with SQL Server 7.0. From a market perspective, Nigel Pendse in The OLAP Report for 2001 states that Microsoft shares OLAP market leadership with Hyperion. Note that Pendse counts by OLAP platform and includes OLAP applications in his market t share assessment. Hyperion’s OLAP business has a substantial application component. Microsoft’s OLAP business is platform only. Therefore, we can infer that Microsoft, in three years, has gone from entry into the OLAP market to become the leading in OLAP platform supplier. And, Pendse predicts that Microsoft will continue to be the fastest growing OLAP supplier.


Within Oracle9i OLAP, Oracle has innovatively implemented a multidimensional storage model through 9i’s Abstract Data Type (ADT) object technology features, implementing OLAP query and analytic processing functions within the database, providing access to those functions through a set of programming interfaces, and integrating OLAP resources within its metadata and management frameworks. Oracle9i OLAP’s storage management and interfaces appear to be brand new. Its OLAP functionality appears to leverage the functionality of Express.

Within Oracle9i, OLAP resources theoretically can be managed with the same tools as relational and object resources. OLAP, relational, and object resources are made scalable, reliable, secure, and manageable by the same set of mechanisms. That’s an advantage. The disadvantage is lack of maturity. This approach is unproven. In addition, the current release requires separate relational and OLAP build and manage toolsets.

Oracle’s logical OLAP data model includes the familiar dimensions and measures. It can be defined within and/or across relational and multidimensional storage, enabling Oracle’s business intelligence platform to implement MOLAP, ROLAP, and HOLAP storage models. Analytic processing capabilities are quite rich. In addition to handling the general OLAP aggregation, allocation, and multidimensional navigation functionality, 9i OLAP also packages financial calculations and functions, statistical functions and statistical forecasting, and regression. Both Microsoft and Hyperion also package good numeric analytic functionality, but forecasting is Oracle’s differentiator in analytic functionality. Forecasting was also a differentiator of Express.

Oracle’s Fourth Try at OLAP

Oracle9i Release 2 OLAP is Oracle’s fourth approach to OLAP technology. The first was through the 1995 acquisition of Information Resources, Inc. (IRI) and that firm’s Express OLAP technology. Express, a market leader at the time of the acquisition, was never integrated into Oracle’s products and never seemed to be a strategic element of Oracle’s data warehousing or business intelligence strategy. It seemed to whither away from a lack of R&D and marketing nourishment.

The second OLAP attempt came with the introduction of Oracle8i in 1999. Oracle felt that 8i’s analytic SQL extensions, performance, query optimization, indexing, and partitioning capabilities would enable it to deliver the performance and usability of a native multidimensional database through a relational store. That didn’t happen.

The third approach was Oracle9i Release 1 OLAP, which was introduced in April 2001. Perhaps, as a typical “1.0” version, OLAP Release 1 was unusable.

With 9i Release 2 OLAP, Oracle has delivered more usable OLAP technology and has begun to demonstrate that it understands that the advantages and benefits of OLAP can be achieved only through a real OLAP implementation. Now the firm must convince customers to use the facility and to do the product marketing work convince application suppliers to make R&D investments in 9i OLAP.

IBM and Hyperion

OLAP capabilities for the IBM business intelligence platform are provided through DB2 OLAP Server, the re-branded Hyperion Essbase. Essbase is widely used, well proven, and has broad support from applications and tools partners. Its OLAP technology and analytic functionality are quite strong. Essbase is in the “point 5” release of its sixth major version, and Hyperion has made major improvements in each. Features that differentiate it from the other OLAP offerings are platform support for IBM mainframes, Unix, and Windows NT as well as wide range of analytic applications from Hyperion and from its ISV partners. On the other hand, Essbase has become less attractive as a dedicated OLAP system as first Microsoft and later Oracle has integrated OLAP capabilities into their relational databases, creating comprehensive business intelligence platforms.

Hybrid Analysis is the best new feature of DB2 OLAP Server V8.1 (and the best new feature of Essbase 6.5 XTD, the current and corresponding newest version from Hyperion. It was introduced in April 2002). Hybrid Analysis allows administrators to define a multidimensional cube that stores its high-level (aggregated or summary) members within a multidimensional store and its low level (detail) members within a relational store. The OLAP Server accesses and prepares the relational data and presents this data to the application as if it were native multidimensional data. Hybrid Analysis insulates OLAP applications from the complexities of relational drill-through, the approach of previous Essbase/DB2 OLAP Server versions. It adds HOLAP (Hybrid OLAP) support to the OLAP storage model of IBM’s business intelligence platform. This is a very attractive and useful feature. Of course, Microsoft Analysis Services has provided a similar capability since its introduction almost four years ago.

DB2 OLAP Server offers very good OLAP capabilities. The new integration of HOLAP storage models addresses a significant limitation with prior product versions. Our issue in OLAP is in integration. DB2 OLAP Server is an add-on to IBM’s business intelligence platform. It requires its own data store, separately managed from the relational data warehousing data store, and its own build and manage toolset. This approach results in adding to the resources and skills that you need to implement and support IBM’s business intelligence platform.

Comparing Data Mining

While businesses have been using data mining technology to good advantage for many years, data mining functionality has only recently been integrated into business intelligence platforms. In the past, data mining technologies analyzed data samples from database and data warehouses. The analysis was limited by the relatively small amounts of data that data mining algorithms could process. It was also limited by the complexity in using data mining technologies. You needed the skills of a statistician and a DBA to be able to use data mining effectively and efficiently. Today, business intelligence tools and applications abstract this complexity to make data mining more accessible and more usable. In addition, data mining technologies are no longer limited by the amount of data that they can process. Their input can be an entire data warehouse.

The business intelligence platforms of Microsoft, Oracle, and IBM all integrate data mining capabilities. We can differentiate these offerings by packaging, tools, implementation, breadth of data mining functions and algorithms, and input. Note that data mining interfaces are also a significant differentiator. We’ll examine interfaces in the next section of this report. Note also that Hyperion does not provide data mining functionality.

Microsoft Corporation

In SQL Server 2000, Microsoft added data mining capabilities to the OLAP functionality packaged in SQL Server Analysis Services. This is a good first step for Microsoft in this sophisticated area of business intelligence analysis. The best features are packaging, integration within Microsoft’s business intelligence platform, wizard-driven model building, and the capability to mine relational, OLAP, or external OLE DB data. On the other hand, Analysis Services implements only two data mining algorithms: decision trees and clustering. While these algorithms can be applied to address many prediction problems, additional algorithms would add flexibility and broaden the range of problems that could be solved.

Note that Analysis Services supports the Predictive Model Markup Language (PMML) specification for the interchange of data mining models. IBM Intelligent Miner and Oracle9i Data Mining support PMML, too.


Data mining capabilities of Oracle’s business intelligence platform are delivered through Oracle 9i Data Mining, a separately packaged and priced feature of Oracle9i. New in Oracle9i Release 2, this feature is based on the Darwin technology that Oracle acquired from Thinking Machines Corporation in June 1999. Its classification, clustering, association rules, and attribute importance data mining functions are implemented by adaptive Bayes Network, Naive Bayes, k-Means, O-Cluster, predictive variance, and A priori algorithms.

Oracle9i Data Mining gives Oracle a good start in providing data mining capabilities within its business intelligence platform. The best features are the integration within the Oracle9i database, the use of the database for input and metadata, and the broad range of functions and algorithms. On the other hand, Oracle9i Data Mining is obviously a new offering. Data visualization functionality is missing. There’s no support for a data mining process. Visual modeling tools are also missing. And, there’s little support from business intelligence tools and application suppliers.


IBM has been active in data mining much longer than Microsoft or Oracle. Early versions of its DB2 Intelligent Miner for Data, then called simply Intelligent Miner, were among the early data mining workbenches. For DB2 Version 7 IBM integrated Intelligent Miner into its DB2-based business intelligence platform as DB2 Intelligent Miner for Data. The product has three components:

  • DB2 Intelligent Miner Visualization

  • DB2 Intelligent Miner Modeling

  • DB2 Intelligent Miner Scoring

Each supports a key phase of the traditional data mining process. All three are separately priced and packaged. The strengths of DB2 Intelligent Miner are the breadth and depth of its data mining algorithms and its modular and comprehensive coverage of the data mining process. Its limitations are complexity—the tradeoff for its power and flexibility, and support only for relational input. Regarding complexity, IBM has not abstracted data mining within higher-level, easier to use analytic applications. Not many business intelligence tools and applications suppliers have either.

For OLAP data mining, DB2 OLAP Server includes DB2 OLAP Miner. DB2 OLAP Server Miner applies data mining technology to OLAP. More specifically, it uses a single statistical algorithm to discover cells within an OLAP dimension that have unexpected or outlying values and presents them visually to analysts. Analysts can then use traditional OLAP approaches to determine and to understand the causes for value differences. This is a very useful data mining application. It automates an important business intelligence process and can make the results of that process consistent. Also, unlike Intelligent Miner, OLAP Server Miner does not require skill and experience in data mining. OLAP experience is all that you need to use it. This is the right approach to data mining applications.

DB2 OLAP Server Miner was introduced in November 2001. This product was developed by IBM. It’s the first and so far the only OLAP component that IBM hasn’t OEMed from Hyperion.

Individually, these products provide good data mining functionality. Together, they offer broader and deeper data mining functionality than either Microsoft or Oracle, but they’d be stronger if they were integrated into a single coherent offering. It’s a disadvantage that Intelligent Miner doesn’t mine OLAP data. It’s also a disadvantage that DB2 OLAP Miner has only a single data mining algorithm.

Comparing Interfaces

Interfaces are the mechanisms for accessing data and functionality of business intelligence platforms by business intelligence tools and packaged applications. Open, usable, flexible, and adaptable interfaces make a business intelligence platform more attractive to tools and applications suppliers, giving you more choice in complementing your business intelligence platform. They also make it easier for you to create your own business intelligence applications. The interfaces for relational data, OLAP, and data mining for each of the four leading business intelligence suppliers are listed in Table 1.






Relational interfaces

SQL and Transact/SQL





    SQL and PL/SQL

    ODBC and JDBC



Not applicable

OLAP Interfaces



    Pivot Table Service

    XML for Analysis




Essbase API

Essbase API

Data mining interfaces


Pivot Table Service


Oracle9i Data Mining API (Java)

Intelligent Miner

  • C++

  • SQL

  • Visual tools

DB2 OLAP Miner

  • Essbase API

Not applicable

Table 1: This table lists and describes the relational, OLAP, and data mining interfaces within the business intelligence platforms of Microsoft, Oracle, IBM, and Hyperion.

Microsoft Corporation

Relational Interfaces

Microsoft provides the broadest range of relational interfaces. These interfaces provide good flexibility, supporting a range of business intelligence application types and developers. SQL and Transact/SQL, SQL Server’s SQL dialect, are conventional procedural SQL DML interfaces. OLE DB and the ADO interfaces are COM interfaces that encapsulate SQL and Transact/SQL DML within rich, object-oriented programming structures. They’re especially usable, flexible, and adaptable. ADO.NET, the newest relational interface is an especially attractive interface for distributed, Web-based applications. Here’s a little more detail on Microsoft’s relational interfaces:

  • ODBC and JDBC are the lowest level application interfaces to SQL Server relational data. Both define a procedural call-level interface. SQL Server supports ODBC and JDBC as native APIs. These are the interfaces for applications that don’t use COM.

  • OLE DB provides a higher-level object-oriented application interface. Microsoft recommends OLE DB for developing tools, utilities, and components, where high performance and flexible data manipulation are required. SQL Server packages native OLE DB support in its OLE DB Provider for SQL Server.

  • ADO (ActiveX Data Objects) is the highest-level application interface to SQL Server relational data. ADO encapsulates and abstracts OLE DB, providing object-oriented facilities to connect to, retrieve, manipulate, and update data from an instance of SQL Server. ADO insulates application developers from the complexity of programming COM interfaces. Microsoft recommends ADO for general-purpose access programs in business applications.

  • ADO.NET is an ADO interface that Microsoft has designed for access to data from remote, Web-based applications. The structure of its components and their recommended usage result in minimizing network roundtrips between the application and the database. ADO.NET applications access the database in short bursts, connecting for only the time needed to perform a single database operation and caching relatively large amounts of data.

OLAP Interfaces

Analysis Services provides four OLAP (and data mining) application interfaces:

  • MDX (MultiDimensional Expressions). MDX is Microsoft’s native OLAP interface. It’s the interface to Analysis Service’s multidimensional data that’s analogous to the SQL interface to relational data. Like SQL, MDX provides both data definition syntax and data manipulation syntax. Like SQL queries, each MDX query has a data request (the SELECT clause) that may be qualified with a starting point (the FROM clause) and a filter (the WHERE clause). Analysis Services defines over 100 MDX functions. Applications use MDX data manipulation functions within the DSO, PivotTable Service, and XML/A programming models.

  • DSO. Decision Support Objects (DSO) defines a COM-based object model that provides an interface to the internal structure Analysis Services OLAP and data mining functionality and data. Its objects encapsulate server platforms, SQL Server databases, MDX functions, OLAP data structures, data mining models, and user roles.

  • PivotTable Service. PivotTable Service is a client-based OLE DB provider for Analysis Services OLAP and data mining functionality. Through the OLE DB object model, PivotTable Service applications can access, manipulate, and retrieve relational and multi-dimensional data, create local multidimensional cubes on the client, perform OLAP and/or data mining functions on those cubes, and can display the results of all its processing. This is a powerful but heavy client interface that can be very attractive to mobile applications.

  • XML for Analysis (XML/A). XML for Analysis is a Simple Object Access Protocol (SOAP)-based XML API that has been designed by Microsoft for accessing SQL Server Analysis Services data and functionality from Web client applications. It uses these key, standard, Web services protocols to create an OLAP interface that is both language independent and requires no pre-installed client components. XML/A eliminates hardware platform, operating system, user interface model, programming language, and version dependencies both on the client and between the client and Analysis Services. In addition, its runtime characteristics for database resource allocation are tuned for the processing style of Web applications.

These interfaces can support all types of business intelligence applications. PivotTable Service and XML/A are especially attractive. PivotTable Service is a nice fit for occasionally connected mobile applications. XML/A implements the ideal architecture for Web applications. It transforms SQL Server Analysis Services into a Web Services provider for OLAP or data mining. None of the other suppliers even approaches providing interfaces that are so closely tailored for supporting types and processing styles of OLAP applications.

Data Mining Interfaces

For building data mining models Analysis Services provides a wizard-based interface. Alternatively, models can be built programmatically using the DSO object model. The PivotTable Service also provides capabilities for building and training data mining models. DSO and PivotTable Service are the interfaces for using data mining models to score OLAP or relational data.

Microsoft’s data mining interfaces have these advantages. Visual tools such as the Data Mining wizard simplify the complex model building process. PivotTable support is also a strong feature of Analysis Services data mining interfaces. There are some disadvantages of these interfaces, too. For example, we’d like to see better data visualization within the data mining input process.



Relational Interfaces

Oracle’s SQL and PL/SQL relational interfaces are the traditional procedural interfaces that it has long supported. PL/SQL is widely implemented by business intelligence tools and applications. Oracle doesn’t have the higher level, more flexible interfaces like Microsoft OLE DB or ADO.

OLAP Interfaces

Oracle defines three interfaces to Oracle 9i OLAP:


  • Java OLAP API

  • SQL and PL/SQL

OLAP DML is a procedural language that’s the most flexible and most functionally comprehensive interface to Oracle9i OLAP data and analytic functions. Think of it as the native interface to Oracle9i OLAP. Through OLAP DML, applications can access, query, navigate, and manipulate multidimensional data as well as perform analytic functions. All of Oracle9i OLAP’s analytic functionality is accessible from OLAP DML and OLAP DML can be used to define new analytic functionality.

Through the Java OLAP API, applications can connect to multidimensional data and can perform navigation, selection, analysis functions, and cursor management functions. However, not all analytic functionality is accessible through the Java OLAP API. Forecasts, for example are not. Java applications must execute OLAP DML commands when functionality is not available through the Java OLAP API. That’s a disadvantage.

Oracle9i OLAP provides three ways to access OLAP data and functionality through SQL. First, predefined PL/SQL packages access OLAP DML commands directly. Second, additional predefined PL/SQL packages use table functions to create views of OLAP multidimensional data. SQL applications can access these views. Third, SQL applications can access Oracle9i OLAP table functions directly, creating their own views.

There’s good flexibility in these interfaces. The power of OLAP DML and its capabilities to extend analytic functionality are significant advantages. So is the capability to access Oracle9i OLAP data and analytic functionality from SQL. That can leverage investments in existing business intelligence applications. On the other hand, the incompleteness of the Java OLAP API is a disadvantage, hopefully a temporary one. Also, these are all complex interfaces. It won’t be an easy job to extend SQL applications for OLAP access, but it is an advantage that OLAP access is an extension not a re-write. Higher level interfaces with more abstraction would be quite helpful. The PL/SQL packages are a good start in that direction.

Oracle9i OLAP interfaces are very new. Few, if any business intelligence applications have integrated them. Only a few business intelligence application suppliers have committed to make the R&D investments necessary to support them. Not even Oracle’s business intelligence applications support them yet.

Data Mining Interfaces

Oracle9i Data Mining provides a Java API for accessing its data preparation, modeling, testing, and scoring functionality. This is an open API and Oracle makes its published specification easily available. Note though, similarly to the Oracle9i OLAP interfaces, that there are few, if any, business intelligence applications that use the Oralce9i Data Mining API or the functionality available through it, not even Oracle’s business intelligence tools and applications.

IBM and Hyperion

Relational Interfaces

In a manner similar to Oracle’s relational interfaces, IBM’s SQL and DB2 SQL relational interfaces are the traditional procedural interfaces that it has long supported. Also, like Oracle, IBM doesn’t have the higher level, more flexible interfaces like Microsoft OLE DB or ADO.

OLAP Interfaces

Because DB2 OLAP Server is IBM’s re-branding of Essbase, it implements the Hyperion Essbase Application Programming Interface (API) to data access and OLAP analytic function access. This interface provides access to OLAP data and analytic functionality through C, Java, and Visual Basic bindings. This is the interface that business intelligence tools and application suppliers use to integrate Essbase within their product offerings. It’s also the interface used by Hyperion for its business intelligence tools and applications as well as for its administrative tools. The Essbase API is an open interface. Its published specification is readily available. It is widely supported by Hyperion’s partners.

Data Mining Interfaces

Data mining interfaces within the IBM business intelligence platform are open and support appropriate standards. However, few business intelligence tools and applications support Intelligent Miner. It’s most commonly used as a data mining workbench. DB2 OLAP Miner is a packaged application that uses the Essbase API to access and retrieve the data to be mined. It present the results of data mining through the user interface. This approach makes its data mining functionality easily accessible to OLAP applications.

Comparing Build and Manage Capabilities

Build and manage capabilities are the toolsets and mechanisms that you use to create relational and multidimensional data warehouses and data marts, to populate them with data, and to control their content and usage. Build and manage capabilities differentiate business intelligence platforms in the areas of the number, integration and packaging of toolsets, the range of data sources supported for data warehouse and data mart input, and how extraction, transformation, and loading are implemented and executed. These aspects of the build and manage capabilities of the four leading business intelligence suppliers are listed in Table 2. Our analysis and comparison follows in the sections below.


Microsoft has significant advantages over competing suppliers in build and manage toolsets. Their packaging, pricing, and integration are key strengths of Microsoft’s business intelligence platform and a major competitive advantage. Oracle and IBM offer toolsets with user interfaces that aren’t as visually rich or highly functional. In addition, some of their higher value capabilities are packaged as extra-cost options.

Microsoft’s build and manage functionality also has other strengths and advantages.

  • The process and workflow orientation of DTS packages is flexible and easily adaptable. It can automate ETL execution. Also, transformation capabilities can be easily extended using object-oriented techniques on their COM implementation. Oracle also provides process-oriented build and manage capabilities, but IBM and Hyperion don’t.

  • DTS support for non-database data structures and for files is particularly strong.

  • DTS has integration facilities that organize its packages into transactions. They execute completely or their effects are rolled back. Transactional Lookup queries can incorporate data not in the DTS source connection into the transformation task. Lookup queries might be a way to augment name and address information.

  • DTS provides management capabilities on its packages. They can be versioned and password protected. This is a key strength.

  • DTS packages execute on the Windows server platform outside the database. Their processing does not interfere with the processing of production business intelligence applications. This approach contrasts with the transformation approach of IBM and Oracle, which use processing facilities of their database engines for transformation.

Build and Manage Capabilities






Analysis Manager provides comprehensive relational and OLAP build and manage capabilities.

    Oracle9i Warehouse Builder provides relational build and manage capabilities.

    Oracle Enterprise Manager provides OLAP build and manage capabilities.

DB2 UDB Data Warehouse Center (DWC) provides basic relational basic build and manage capabilities.

DB2 Warehouse Manager adds additional relational build and manage capabilities.

DB2 OLAP Administrative Services provides OLAP build and manage capabilities.

Essbase Administration Services provide OLAP build and manage capabilities.

Integration Server provides support for loading and accessing relational data.

Extraction data sources

    Microsoft SQL Server




    Access 2000, Excel 2000

    Microsoft Visual FoxPro

    dBase, Paradox

    Microsoft Exchange Server

    Microsoft Active Directory



Microsoft SQL Server







Microsoft SQL Server






Microsoft SQL/Server Oracle



Additional Extraction Data Sources

Host Integration Server provides extraction from IBM mainframe data sources.

Oracle Pure Extract provides extraction from IBM mainframe data sources.

Oracle Warehouse Builder Integrator for SAP provides extraction from SAP R/3.

DB2 Warehouse Manager provides extraction from SAP R/3, i2, and Web Server logs.

Tools from IBM partners ETI and Ascential integrate within DWC to provide additional ETL capabilities.


ETL execution

Process-oriented execution of tasks within packages. Packaged may be versioned and/or password protected.

Process-oriented and execution of ETL steps controlled by Enterprise Manager.

Individually executed ETL steps.

Procedural sequences of declarative rules.

ETL implementation

DTS is implemented as a COM framework accessed programmatically or with packaged visual tools.

PL/SQL stored procedures in Oracle9i database,

DB2 stored procedures and user defined functions (UDF). 150 predefined transformations.

ETL performed through rules. Rules perform field-level operations on source data. A set of predefined rules is packaged.

Data cleansing

None packaged.

Oracle Pure Name and Address provides name and address data cleansing.

IBM partners Trillium Software Systems provides name and address data cleansing.

Via user-defined rules.

Table 2: These tables lists and describes the build and manage capabilities within the business intelligence platforms of Microsoft, Oracle, IBM, and Hyperion.

On the other hand, there are also a few disadvantages. For data source s, we’d like to see more natively supported databases and support for the data of the popular application packages from suppliers such as SAP and PeopleSoft from within the databases that those package support. This type of support helps business intelligence requirements in the largest companies. Note that competing approaches from IBM and Oracle have this limitation, too. In addition, missing in DTS, as it’s missing in competing approaches, are packaged data cleansing functions and name and address merge, purge, and de-dupe functions. These missing functions can be provided by integrating products from Microsoft’s partners. In contrast, Oracle provides a separately packaged and priced product, Oracle Pure Name and Address and IBM provides a separately packaged and priced but integrated partner product from Trillium.


When you implement Oracle9i relational and multidimensional data warehouses, you’ll need both Enterprise Manager and Warehouse Builder. Additional Oracle tools and products will most likely also e necessary. The number of products required and their separate pricing and packaging is a disadvantage of Oracle’s build and manage capabilities.

Oracle’s build and manage functionality does have significant advantages. Warehouse Builder has very good extraction capabilities, although they’d be better if supported data sources included the data of other popular application packages in addition to SAP R/3. Warehouse Builder packages a large set of predefined transformations and provides visual tools for customizing them or for developing new ones, although their stored procedure implementation is not as powerful or flexible as the COM implementation of DTS transformations. Also, like DTS, Oracle Enterprise Manager provides tools for specifying business intelligence platform processes. The processes may implement any and all build and manage steps and Oracle Workflow may be used for scheduling these processes.

The major limitation of Warehouse Builder is that its current version doesn’t provide OLAP build and manage capabilities. That’s done within Oracle9i Enterprise Manager. Also, while implementing transformations within the database has the advantage of simplified management, it’s an approach that centralizes all aspects of data warehousing. Transformation processing can interfere with business intelligence application processing when they’re performed concurrently. The database engine can be taxed by attempting to manage these workloads. Installations may not have tuned their databases to manage this processing. In contrast, as we mentioned earlier, Microsoft executes its transformations outside the database, allowing the operating system to balance the workloads.


Data Warehouse Center (DWC) is a basic build and manage facility. Its major advantage is its packaging as a DB2 UDB feature. DWC could be the only build and manage toolset that your administrators need, but only if you have just relational warehouses with conventional data sources and simple transformation requirements. If your business intelligence platform includes OLAP, if it has the popular packaged operational application data sources, and if it requires more sophisticated transformations, then you’ll need additional build and manage products. These additional products are Warehouse Manager for additional data sources and more efficient data movement, DB2 OLAP Administrative Services (The re-branded Hyperion Essbase Administrative Services) for OLAP build and Manage, and, perhaps, ETI and Ascential for more sophisticated transformation. Note that Warehouse Manager packages extraction capabilities from popular software environments: SAP R/3, i2, and the Web (for analyzing Web traffic). This array of separately packaged and priced products puts IBM at a disadvantage for packaging, integration, and complexity, especially so when compared with Microsoft’s build and manage approach.

For the functionality implemented by these toolsets, extraction functionality is implemented within the database and is accessible through SQL. Its transformation functionality is implemented as stored procedures and User-Defined Functions accessible through SQL. The use of SQL makes these capabilities easy to learn, easy to use and easy to manage for DBAs. But, a SQL implementation can limit their richness and power and their execution can cause interference with business intelligence application processing.


Essbase Administration Services is the right toolset for OLAP build and manage when OLAP is part of a comprehensive business intelligence platform. While this toolset has no data cleansing capabilities and its procedural, rules-based extraction and transformation functionality is complex, remember that Essbase is not a complete business intelligence platform. Within a business intelligence platform, the data sources for Essbase are commonly data warehouses, and data warehouse data is already transformed and cleansed. The transformation and cleansing capabilities of the other three business intelligence platforms discussed in this report as well those for specialized build and manage tools are much richer and more flexible than those in Essbase. However, when Essbase data sources are operational systems, additional transformation and cleansing products can be required for successful implementation.


Microsoft, IBM, and Oracle address all of our business intelligence platform requirements. They provide relational data warehousing, build and manage facilities, OLAP, data mining, and application interfaces to relational data warehouses, to OLAP data and analytic functionality, and to data mining. Hyperion provides only OLAP. Summarizing our analysis of each business intelligence platform:

  • Microsoft provides a comprehensive business intelligence platform. Build and manage capabilities, OLAP capabilities, and application interfaces are its key strengths. Data mining is very new, although data mining integration and data mining tools are quite good.

  • Oracle provides a comprehensive business intelligence platform. While this platform has a complete set of components, OLAP and data mining capabilities are unproven, data mining tools are low level, and build and manage capabilities are not consistently implemented for relational and OLAP data.

  • IBM provides a comprehensive business intelligence platform. Relational data warehousing and data mining are key strengths. OLAP capabilities are very good, but because they’re OEMed from Hyperion, the platform is not well integrated. Build and Manage capabilities require too many toolsets and there’s a disconnect between managing relational data and managing OLAP data.

  • Hyperion occupies a niche within business intelligence platforms, but its OLAP technology is a critical component of IBM’s business intelligence platform.

When packaging and pricing are factored into the analysis, Microsoft leaves IBM and Oracle behind. Microsoft’s entire platform is included within SQL Server 2000 Enterprise Edition at a price of $19,999 per processor. For IBM and Oracle, each platform component is separately priced. The per processor price for IBM’s business intelligence platform starts is $138,600 before adding per user fees for OLAP. The per processor price for Oracle’s business intelligence platform is $80,000 before adding $5,000 per user fees for build and manage capabilities. Even with the assumption of equivalent capabilities, Microsoft delivers at least five to eight times the value of IBM or Oracle. But, that’s not a good assumption because Microsoft provides at least equivalent capabilities for most platform components, lesser functionality only in data mining and only compared to IBM, and greater functionality in build and manage, OLAP, and application interfaces.

Download 145.88 Kb.
leave a comment
Date conversion21.10.2012
Size145.88 Kb.
TypeДокументы, Educational materials
Add document to your blog or website

Be the first user to rate this..
Your rate:
Place this button on your site:

The database is protected by copyright ©exdat 2000-2017
При копировании материала укажите ссылку
send message