Intelligent Enterprise

Better Insight for Business Decisions

RSS
Webcasts
Whitepapers
Subscribe
Home

May 31, 2003

Oracle Drill Down

Already a dominant player in data management, Oracle is on a quest to deliver the business intelligence "stack" as a tightly integrated whole. Is the sum greater than the parts for your business analysis needs?

by Jack Hakim and Tom Spitzer

Continued from Page 2

The power of the BI Beans approach to analytic applications stems from its enabling an object-oriented model for defining queries and calculations that can go arbitrarily deep into the internal OLAP API. Also, BI Beans has the ability to create completely custom user interface models and wire them to the queries and calculations. Several third-party companies are leveraging this power to build their own end-user BI tools, including Vlamis Software Solutions and IAFSoft. (For a closer look at how Vlamis's VSS Business Analyzer uses BI Beans, see Product Review.)

OLAP Services

The strength of the Oracle BI offering lies in the tight synchronization of BI Beans with Oracle's OLAP API. Put together with the analytic engine that now resides inside the Oracle9i database (formerly delivered as the stand-alone Oracle Express), and you have Oracle OLAP services. OLAP services can do complex statistical, mathematical, and financial calculations, along with predictive analysis such as forecasting, modeling, consolidation, allocation, and scenario management.

The Java OLAP API lets developers work directly with a multidimensional object model. It also offers a complete set of OLAP calculation functions and support for asymmetric queries such as nested rankings. Queries issued through the BI Bean use the Java OLAP API, which can then determine whether the data required to fulfill the request resides in a relational database or analytic workspace. If stored in a relational database, the API generates SQL to represent the query, which can include things like N-pass functions, ranking, and a variety of aggregates. If data resides in an analytic workspace, the API can generate a request in OLAP data manipulation language (DML) — the native data retrieval language — to submit the request to the analytic engine.

OLAP services' SQL table functions can take a set of rows as input and produce a set of rows as output, which users can query like a physical database table. Oracle provides PL/SQL packages that use table functions to create views of multidimensional data residing in an analytic workspace. SQL applications can access these views. Thus, the calculation engine and analytic workspace data are accessible to SQL, making analytic and predictive functions available to SQL-based applications. In addition to using PL/SQL procedures for accessing analytic workspace data as SQL views, application programmers can use the Oracle OLAP packages to execute OLAP DML commands directly and return the results to their applications.

The Oracle calculation engine understands OLAP DML, which thus extends the analytical capabilities of querying languages such as SQL and the OLAP API to include forecasting, modeling, and what-if scenarios. Application developers can create stored procedures that use conditional logic, as well as the extensive library of DML commands and functions to perform complex data analysis. Moreover, the OLAP DML provides a calculation language, similar to that of a spreadsheet, which is easy for power users and DBAs to learn and use.

How Does It Play?

We were a little surprised to learn how coherent and comprehensive Oracle's BI story has become, and how easy it is to take advantage of the basic BI functionality. Oracle's tight database integration should, in the long run, enable optimized performance for analyzing rapidly increasing volumes of data while maintaining database reliability and security. This secure scalability has become increasingly important in the Internet age.

Users and developers' main concerns had to do with performance, administration, and living in the real (that is, heterogeneous) world. They were also concerned about finding a balance between simplicity and complexity that allows everyone to just get the job done. For example, some Oracle customers have deployed Discoverer for simple queries but continue to use competing products, such as Business Objects, which they view as offering more powerful report-building capabilities without having to make the leap to the Java-based tools for creating the analytic reports provided by BI Beans.

Corporate hesitancy to standardize exclusively on the Oracle BI stack comes from a reality that many face: a multidatabase environment. Oracle offers gateway capabilities. However, Informatica and other data integration middleware specialists that knit together analytic solutions for multidatabase environments still make points by emphasizing their expertise in consolidating metadata from disparate systems, optimizing query performance in heterogeneous environments — and appealing to corporate concerns about the risk of vendor lock-in.

Great differences exist between the offerings of Oracle, Informatica, and other vendors in terms of visual metaphors, integration issues, and cost. A fair comparison would be the subject of another article. Also, an emerging option to watch is Web services, which could bridge point solutions from multiple vendors. However, we have yet to see a significant number of deployments that deliver reliability, security, and scalability at cost points organizations desire.

Rate This Article

Comments:

Optional e-mail address:

Oracle has made it possible to do almost any business analysis you want at a reasonable price. But if your needs require customization, serious Java and SQL development remain a big part of the solution.

Jack Hakim [jhakim@ecwise.com] and Tom Spitzer [tspitzer@ecwise.com] are principals at EC Wise Inc., with offices in the San Francisco Bay area and Seattle.

RESOURCES

Oracle: www.oracle.com

Related Articles at IntelligentEnterprise.com:

"The Oracle Way," March 1, 2003