Defining IT Information

Given the complexity of enterprises and the rate of change in today’s business climate, consistently representing EA information is crucial because such a representation serves as the vocabulary that describes the enterprise and, thereby, provides the capability to align IT and the business. This problem is typically addressed by developing a meta model that describes the objects and relationships in the enterprise.

The meta model must allow for the description of not only classes of objects, but also the classes of relationships possible between those objects. A meta model must be rich enough and extensive enough to capture the salient attributes of both objects and relationships. Further, the modeling capability must have the flexibility to define both the conceptual/intangible objects (e.g. a business process) and tangible/physical entities (e.g. a server). To address future needs, the meta model must be extensible and scalable so that changes to meta model definitions can be inherited by the existing information in the MDR.

The ability to define a meta model is necessary, but it is not sufficient to meet the organisational goals. To meet the goals of actionable EA, the model must be common and consistent across all IT goals and processes so that all roles use this same meta model and therefore leverage an integrated view of the information. Further, the meta model must be designed to support views, queries, analytics, and policy application necessary for actionable EA. Such a meta model provides the syntax and semantics (the grammar), necessary to implement BI for IT. The meta model represents the breadth and depth that the MDR is capable of representing.

Acquiring IT Information

Once the reference model is designed, it needs to be cost-effectively populated with current, accurate, and complete information that is stored in an MDR. Without the population of the MDR with all the necessary data, the promise of EA will fail. To this end, the data must be acquired in the most automated means possible from the authoritative sources in the enterprise. In addition, the automated collection of the information about the entities is insufficient - the automated discovery of information about the entity relationships based on rules is also a necessity.

Most enterprises have a large number of information sources (e.g. silos) that need to be aggregated into a strategic information MDR. Therefore, the platform must support a full suite of Extract Transform and Load (ETL) integrations with those sources to facilitate the seamless integration of this information into the MDR.

In order to minimise the costs of aggregation and deployment, the platform should not require modification of the sources, nor should it require “agents” to be deployed and installed in the source environments. In addition, because every enterprise has some non-standard sources, a toolkit must be available to enable custom ETL collections in a cost-effective manner. Further, the platform must be capable of regularly scheduled refreshes of the data based on the volatility of each source.

To be scalable, the ETL integrations must support rules that will resolve any conflicts in names, as well as define precedence when multiple sources produce conflicting values for the same attribute. Ideally, the ETL integrations should enable the ability to configure business logic unique to the enterprise. 

Subject matter experts represent key information sources and they exist in a variety of roles all of which must be supported. Some roles will require easy-to-use web interfaces that are broadly available. Others will require powerful visual modeling, analytic and presentation capabilities that deliver integrated text and graphic-based models. The ability to support collaborative teams, including versioning, security, and locking on behalf of subject matter experts is a requirement for any enterprise. Lastly, the amount of information to describe an enterprise is vast. The platform must be capable of handling millions of entities, relationships, and associated attributes.