Oracle OBIEE 11g: The Stuff of Semantic Dreams?

Oracle's launch of Oracle Business Intelligence Enterprise Edition (OBIEE) 11g sent a strong signal that the company is finally ready to rumble in the BI market. Oracle says that the development team, drawn from its stable of homegrown and acquired talent, is now unified; instead of a three-year gap between releases, it is promising continuous, incremental enhancements, thanks to adoption of Agile development methods. But while a steady stream of new features and functions is important, perhaps even more so is Oracle's focus on improving its underlying integration layer to enable users to work with consistent data about customers, products and other objects of interest as they move between tools, applications and data sources. To make BI prosper across its large and sprawling customer base, it needs this layer.

Oracle describes its Common Enterprise Information Model as a "semantic" layer that provides developers with a single resource for data element definitions for use not just with OBIEE 11g but also Essbase OLAP server and Oracle applications. "It is the secret sauce for making BI consistent across all data sources, calculation engines and workflow," said Oracle President Charles Phillips during the July 7 OBIEE 11g launch. Enabling users to access both relational and hierarchical data stored in Essbase (as well as XML data) in an integrated fashion is one of the key advances in OBIEE 11g.Of course, Oracle is not alone in talking up development of such a resource. Enterprise BI "stack" vendors SAP Business Objects (Universe) and IBM (Cognos 8 Framework Manager) are focused on this as well. Other BI and data warehouse system providers, such as Information Builders, Microsoft, MicroStrategy and SAS supply metadata services, layers and master file description sources.

While never perfect, these central data definition resources are a cornerstone of most enterprise BI architectures. They make it possible for developers to work more efficiently as they write queries aimed at multiple data sources, and for end users to exploit self-service access and reporting. Making layers compatible across heterogeneous stacks and BI products is still a work in progress; Oracle's layer currently works within the Oracle environment only. As organizations grow more reliant on metadata or semantic layers, pressure will be brought to bear on vendors to make their layers fit together.

Semantic layers are about more than just providing centralized metadata for data access, however. This is where business concepts and meanings are mapped to the data sources -- that is, where the "semantic" context can guide data access and analysis. Building a semantic layer is an important objective of most master data management (MDM) approaches as well, which means that at least the master data element definitions have to be accepted by more than just BI systems. Emerging business modeling systems such as Kalido Business Information Modeler or Quantrix Modeler could also make use of a semantic layer to enable end users to model higher-level entities such as customers and products rather than get their hands dirty with the underlying complexity of modeling data access and relationships.

Oracle's 11g integration theme may be best expressed in the introduction of Action Framework, which embodies the concept of "actionable intelligence" by allowing users to invoke actions from their BI dashboards. The "actions" could be business processes or Web services. New options for key performance indicators, data visualization and scorecards will help users draw appropriate conclusions from their analysis before launching an action. Oracle's semantic layer will be critical to integrating Action Framework with the range of analytic data resources as well as Oracle applications and business process management systems and service-oriented architecture. To prevent end users from running amok with Action Frameworks, organizations will set up parameters and likely build intelligence into their semantic layers to guide information use and govern privacy, security and more.

Thus, while advances in BI tools and interfaces draw attention, keep an eye on what's going on down in the semantic layer. What vendors and user organizations build there will be the foundation for the next phase of innovation in BI, analytics and intelligent business processes and services.Semantic layers are about more than just providing centralized metadata for data access... Building a semantic layer is an important objective of most master data management (MDM) approaches as well, which means that the master data element definitions have to be accepted by more than just BI systems.