BI Meets EAI: Smart Synergy

If this is the wave of the future for the new real-time enterprise, how do you get there? It's really a matter of understanding your assets in detail and how those assets can be leveraged to run the business better. The process ultimately has four steps:

Note: I'm setting aside the creation of the infrastructure for now, such as the installation of adapters and translators, but you can assume that those need to be included as well.

First you must understand the information sources in detail, both those that are within your control, such as that old ISAM system that won't go away but is nevertheless valuable, and those that are vendor products, such as SAP systems. These systems all store and produce information, so you must understand how that is done, how the information is presented outside of the system, and the meaning of that information. In the case of SAP, which provides almost no visibility into its database, information is presented through the documents that are producible through an API. Some other systems may present a hierarchy, while others send you XML or give you direct access to the data (a rare, but very desirable scenario). As you can imagine, this is no easy task, but it's absolutely critical.

Once you understand the nature of the source systems, you need to identify the metadata and map all information and services found in the source systems to a common logical and physical metadata layer. This is required for the next step, creating abstractions, because you must know what to abstract.

By creating the abstractions, in essence you are recasting the physical representations of the information found in the source systems into a single unified view that better represents the information, making it more useful for analytics. Remember, you are doing this for real-time, operational, and historical information. Here you use middleware products such as IIS technology, integration servers, application servers, and federated middleware, depending on your environment.

Finally, create the analytics to both monitor and analyze the abstracted information. The types of analytics and tools you use depend on the needs of your business. Simple businesses may only need rudimentary monitoring tools with rudimentary statistical capabilities. More complex businesses, on the other hand, may need more complex and powerful tools with valuable features such as forecasting and regression analysis. Once the abstracted real-time and historical information is there, you should be able to swap analysis tools as needed, and even use more that one tool at a time.

The New Synergy

The application integration and BI worlds continue to differ, although they have common technology requirements, solution patterns, and objectives. It's just a matter of finding the synergy and creating the killer technology to get you there.

This technology will provide great value if applied properly within the enterprise. Organizations will have the ultimate visibility into their businesses. All aspects of the business will not only be understood but also situated in the context of the business history. We'll be able to replenish inventories automatically based on past customer behaviors as well as current activity. We'll understand how customers behave during incentive programs, which suppliers provide the best service, what the current trends in productivity are — all data points that, when understood, help business leaders adjust as needed.

David S. Linthicum [[email protected]] is the former CTO of Mercator Software and SAGA Software, both EAI technology companies. He's also the author of the best-selling book Enterprise Application Integration (Addison-Wesley, 1999), which has been translated into 17 languages. His latest book is Next Generation Application Integration: From Simple Information to Web Services (Addison-Wesley, 2003).

Using Ontologies to Link BI and EAI

Most application integration occurs at the information level. You must therefore always deal with semantics and how to describe semantics relative to a multitude of information systems. You also need to formalize this process, putting some additional methodology and technology behind the management of metadata, as well as the relationships therein.

To this end, many in the world of application integration and BI have begun to adopt the notion of ontology. Ontology is a term borrowed from philosophy that refers to the science of describing the kinds of entities in the world and how they are related. Ontologies are important to application integration and BI solutions because they give us a shared and common understanding of the data (and, in some cases, services and processes) that exists within an application integration problem domain. They also help facilitate communication between people and information systems.

By leveraging the ontology concept, you can organize, manage, and share enterprise information, content, and knowledge, providing better interoperability and integration of information systems within and among companies. You can also layer common ontologies within domains with repeatable patterns.

The view of ontologies was best summarized by Quine, who claimed that ontology asks, "What is there?" The answer: everything. In the context of integration and BI, each information system is regarded as a theory that recognizes the existence of a set of objects: its own ontology.

At its essence, ontology is a conceptual information model. Ontologies describe things that exist in a problem domain. These include properties, concepts, and rules, and how they relate one to another, supporting a standard reference model for information integration (the link to application integration) as well as analysis. Ontologies are useful in the science of application integration and BI because they support the human understanding of information. This use is self-explanatory within the context of application integration.

Ontologies facilitate information-based access and information integration across very different information systems. You achieve this by formalizing the application semantics between intra- and interorganizational information resources.