March 1, 2003

The Oracle Way

The company claims BI is a key area for consolidating functionality around its solutions, but does it deliver?

by Jack Hakim and Tom Spitzer

We're living in interesting times. Over the last few years, we've watched the IT industry's focus move from quality, to speed, to cost reduction. Now, the emphasis is on obtaining productivity and ROI from the massive investments in hardware and software made during the late 1990s and into 2000. This emphasis is reflected in several key IT trends:

• More companies are seeing their problems from a "whole" perspective, rather than as a series of point problems, and want to implement a whole solution. Best-of-breed point solutions have often been difficult to integrate; many companies now prefer to purchase a "good enough" whole solution from a single vendor, rather than spend time and money to integrate point solutions from multiple vendors.
• The Web is increasing the importance of scalability, of reaching more people reliably, and of analyzing rapidly increasing volumes of data.
• Web services promise — but have yet to deliver — more affordable, reliable, secure, and scalable solutions that leverage existing systems and best-of-breed point solutions.
• Competing needs for easy access and collaboration, along with increasing security and traceability requirements, pose a significant challenge for IT organizations.
• Analytic applications are assuming equal importance with operational applications. Significant efforts are underway to create workflows that close the loop among these systems.
• Looming in the near future is the promise and threat of "right-time" (often called real-time) enterprise delivery systems. The early emphasis will be on the three or four processes that provide the biggest, quickest bang for the buck, but we expect the potential of obtaining greater efficiencies from new technologies to accelerate this trend. For example, microtechnology has given rise to chips that incorporate data storage and short-range communications capabilities, cost about a dime, and aren't much larger than a dime. Cheap enough to go into most consumer products, they will revolutionize manufacturing, inventory, and logistics by enabling real-time, closed-loop systems that can autonomously make production, marketing, and merchandising decisions based on up-to-the-minute changes in the supply and demand chain. To support the required speed and sophistication, future business intelligence (BI) offerings may need to reside in the database to access data quickly enough, and they will have to employ high-speed analytic engines that process operational data "in place."

We recently attended the 2002 OracleWorld conference in San Francisco to gain some insights into how the company plans to address these issues. Throughout the conference, Oracle executives chanted the mantra of an all-Oracle solution. Based on what we heard, we have some thoughts about whether Oracle offers you a credible migration path to the future reflected in these trends.

On The Whole

Oracle recognized early that the "whole problem" situation is fairly typical; for the last few years, it has focused on creating an integrated set of tools that provide a complete and at least a good-enough solution set. Customers adopting the Oracle9i Application Server (Oracle9iAS) in conjunction with the Oracle9i database, Oracle claims, will obtain higher levels of reliability, performance, and security, while reducing product maintenance, management, and integration costs.

At OracleWorld, company executives singled out BI as a key area for consolidating functionality around Oracle solutions. They claimed that employing all the BI-related features in Oracle infrastructure will let customers eliminate third-party products they use for data extract, transform, load (ETL) processing, report writing, ad hoc queries and analysis, data mining, and development of custom analytic applications.

However, using Web services to cobble together solutions doesn't remove incompatible and redundant models or redundant data. By creating a consistent multiparty model for customers, suppliers, partners, and employees (the trading community model) within its application suite, Oracle has succeeded in eliminating many of the inefficiencies inherent in integrated point solutions. Just creating a single logical customer database across modules has greatly decreased the complexity and redundancy of the full suite offering.

One of Oracle's primary goals is to package a range of features into each of its two infrastructure product lines, Oracle9i and Oracle9iAS. Another is to make features in one product (for example, the Discoverer ad hoc query tool included in Oracle9iAS) leverage the functionality in another (online analytic processing [OLAP] included in the database). Between the two products, Oracle9i Warehouse Builder (OWB), Oracle OLAP, and Oracle9i data mining are available as add-ons to the database, while Reports Services, Discoverer, and Clickstream Intelligence are included in Oracle 9iAS Enterprise edition. You can use the Report Developer tool within the Oracle9i Developer Suite and use this suite's JDeveloper Java IDE with BI Beans to build custom analytic applications.

If your organization embraces the single-vendor model and uses Oracle Financials as its ERP solution, you can also license one of more than a dozen prebuilt subject-area BI solutions (Customer Intelligence, Financial Intelligence, and Sales Analyzer) built on top of the Oracle BI platform. These solutions typically provide a predefined metalayer that organizes views of subject area information, together with a set of predefined queries and reports.

Oracle's BI offering is extensive, and includes both independent pieces of technology (such as Oracle Data Mining) and functionality that has been incorporated into larger products (such as Discoverer). Much of the technology included in the offering was originally obtained through acquisitions; many pieces were designed to serve different needs, and the pace of their integration into the coherent solution has been uneven. To understand their usefulness in the Oracle toolset, you need to think of them in terms of the needs they're designed to address, as well as of the technical environment in which you plan to use them.

Database Analytic Functions

First, let's take a close look at the analytic features of the Oracle9i database platform.

Release 2 of the Oracle8i database introduced a number of functions that support analytic queries to the SQL engine. Of these, the most widely used are probably the Ranking function, which makes it easy to formulate top-N or bottom-N queries, and period-over-period comparison functions, which permit decision makers to determine absolute or percentage change in key performance indicators between two arbitrary periods. In addition, the ability to create moving window aggregates greatly reduces the complexity of obtaining cumulative sums or averages over discrete time periods, while linear regression and correlation functions allow analysts to use SQL syntax to build models.

The 9i release added more functions to the SQL engine to facilitate processing analytic queries against "standard" relational data. For example, inverse percentile query returns the absolute value of the data point that represents a specified percentile within a distribution of values. Hypothetical Ranks and Distributions lets planners and managers create queries that perform a form of "what if" analysis by determining the rank or percentile value a new entity would have if it were inserted into an existing data set. FIRST/LAST functions return the highest or lowest value of a specified column in a group aggregated in accordance with a GROUP BY clause, so this approach could be used to determine the highest margin product in each product line from a query that looked at a large set of products belonging to various product lines.

• News
• Business Intelligence
• Enterprise Applications
• Information Management
• Performance Management
• Process Management
• White Papers

• About Us
• Advertising Contacts
• Technology Marketing Solutions
• Blogosphere
• Intelligent Enterprise Print Archive
• Newsletters
• Contact Us
• Events
• Awards
• PR Info
• Authors
• TechWebCasts
• Database Magazine Print Archive
• Site Map