Another View | The Tao of architecture
Organizational problems often masquerade as a shifting set of symptoms. That can lead to a proverbial blind men and the elephant situation in which well-intentioned parties seek to solve a problem from different perspectives and thus offer widely differing solutions. In information technology, I have seen that approach result in dueling architectures. To end such confusion requires an understanding of the true nature, or tao, of architecture.
Most discussions of IT architecture quickly become tangled in a plethora of abstractions and frameworks: enterprise architecture, segment architecture, system architecture, solution architecture, software architecture, technical architecture, service architecture, federal enterprise architecture (FEA), Department of Defense Architecture Framework, The Open Group Architecture Framework, and on and on. Adding to this confusion are those silver-tongued consultants who offer architectural snake oil to cure whatever ails you. Help!
Let's clarify this once and for all, beginning with an example in the information management space: Given how much data resides in stand-alone systems, should you create a data warehouse or a federated search environment? Both architectures have their benefits and both can integrate data, so how do you decide which is right for your set of problems?
The answer lies in the following four principles:Know the scope of the problem.
The FEA practice guidance document describes three levels of architecture: enterprise, segment and solution. Although solution is a poor name for the lowest level of architecture because it is overly generic ' I prefer system ' those three abstraction levels can effectively segment the solution set and the problem set. The key is to ensure that you match your capability to the need.Achieve line of sight.
One architectural principle is that form follows function ' architects know what they want to build and why they are building it. They know what is superfluous and what is necessary. That knowledge comes from seeing the interconnections from objective to implementation. A good architecture will make that path clear and explicit. A common mistake is making an activity, instead of an outcome, your objective. An activity, such as standardizing your data, is not a good objective because it is only a means to an end. A better objective is information interoperability.Create multiple views.
The blind men and the elephant proverb is so telling because it contains a truth: Different perspectives exist and influence action. Likewise, when you create an architecture, it is not for a single user group but for multiple stakeholders. Good architecture is described from multiple viewpoints with each view providing answers to its particular audience.Optimize for effect.
By far, optimizing key design characteristics is the hardest and most important part of building an architecture. It is also what distinguishes great architecture from good architecture. You optimize key characteristics that enhance your chances for success: aesthetics, usability, scalability, reliability and so forth. This is where fit and elegance come into play. Great architecture fits its users and organizational context in a comfortable way. Great architectural design considers many elements and provides a balance between difficult trade-offs. Finding the right balance of opposing forces is the most difficult part.
In the case of our information management example, which architecture is the right one depends on knowing how well your choices fit your organization and users. For example, agencies often fail by blindly taking a commercial best practice and trying to apply it to government. Different environments require different architectures to properly balance the trade-offs. When an architecture balances opposing forces and fits an organization, the situation and the problem set, it creates an elegant solution.Daconta ([email protected] acceleratedim.com) is chief technology officer at Accelerated Information Management and former metadata program manager at the Homeland Security Department. His latest book is
Information as Product: How to Deliver the Right Information to the Right Person at the Right Time.