Enterprise architecture demands the collection, definition and management of a wide range of complex sets of information. Users need business-centric information, including business goals, mission statements, vision and direction. They also need more detailed information like business requirements and business process definitions. All this business-centric information must be aligned to the IT it supports in the form of technical metadata, data about systems and technologies. Many organizations will face unique and different sets of standards, regulations, and environmental and technological considerations, meaning that the details needed for each organization will be unique and diverse.
Simply having the definitions and descriptions of all the elements of the enterprise architecture is a good start, but users also need to relate dependent artifacts. What requirements fulfill what parts of a business strategy, or what parts of the information back end are fulfilling what concepts of information movement? And who decides how it all relates?
If we have developed such a rich knowledge base, how can we extract value from it? How can we measure the upstream and downstream impact of a proposed change - resulting from a change in the law, a change in business practices, or a change in technology availability?
The information or metadata an organization produces and stores in such a knowledge base creates three distinct organizational views - a business, an information and a technology view. The business view describes the processes within the organization, the information view describes the data and the technology view describes the applications with underlying topologies.
To define an enterprise architecture, an organization must first capture, manage and integrate all this metadata and its dependencies. Doing so can be difficult because metadata is often maintained in a disjointed manner. But by using a proven, model-driven approach, organizations can capture, manage and unite metadata from the business, information and technology views and successfully define an enterprise architecture.
Steps to an Enterprise Architecture The first step to creating an enterprise architecture is capturing metadata. The metadata can be information about the business, information about information, information about applications or information about systems. Capturing metadata involves documenting accurate definitions and descriptions of the elements that make up the business, information and technology views. A key success factor is the ease of use and availability of the metadata capture tools for all roles involved. Standardization on one set of modeling tools and techniques greatly improves the consistency of the metadata captured, and making tools available to all roles that will capture or add value (documentation) to metadata is critical. The next step is managing metadata dependencies, which is where the real value of enterprise architecture lies. The goal is not merely to capture and categorize information, but to understand how it all relates. During this step, integration of the modeling techniques to a common metamodel becomes essential. Integrating the data collection service (models) with the analytical store (metadata repository) happens along the same lines as data warehouse or business intelligence systems. Models are the online transactional processing (OLTP) systems and the repository is the online analytical processing system, but the key is in the transformation of each piece of metadata into a meaningful intersection. These keys are used to perform essential analytics such as change management, impact analysis, cost and risk assessments, gap analysis between “to-be” and “as-is” architectures, etc. A key success factor is in how integrated the modeling tools can be to the repository, and how much of the dependency tracking can be automated versus performed manually. The final step is the ability to integrate capturing with managing metadata and its dependencies. With an integrated environment for enterprise architecture, organizations can determine how business goals relate to implemented systems, how business rules affect the flow of information, and how technology changes impact the top line or bottom line. A key success factor is capturing and maintaining metadata in a timely, natural and accurate way; otherwise, this valuable knowledge base will age and become suspect. Adopting a natural approach to information and dependency capture, coupled with an integrated environment, ensures that the enterprise architecture project develops a knowledge base in a timely and accurate, and therefore reliable, way. The most suitable user interface for this natural capture of metadata is models. There are many different types of models to support the views of an organization; all are easy to use, easy to understand, and for the most part, already a first-class citizen in our organization’s daily lives. Due to the diversity of views that must be supported, there is no single model that will solve the needs of every participant in an enterprise architecture. The Key Role of the Repository Since enterprise architecture spans multiple disciplines and involves multiple perspectives, many different types of models are needed to complete the entire picture. Because of this diversity and the need to support a natural capture, a model-driven approach must support both nongraphical and graphical modeling paradigms. Typically, nongraphical models are used to capture business information, such as goals, strategies, risk and requirements. Graphical models capture business processes, data flows, an application’s structure and behavior, and the information architecture from both structured and unstructured data sources. However, simply having an integrated modeling toolset that can capture and maintain multiple separate and distinct models with their related dependencies is not enough; this metadata must be managed with a nonproprietary, robust and integrated repository - resulting in an integrated modeling environment.
