Conceptually, enterprise architecture is a logical framework in which to align business intent with information technology assets. Decomposing a problem domain into smaller, yet more manageable sub-problems is an age-old solution approach. Indeed, most enterprise architecture solution methodologies decompose the enterprise (i.e., the problem domain) into some combination of business, application, data and infrastructure architectures (i.e., sub-problem domains).
However, in practice organizations struggle to realize value from the enterprise architecture exercise for several reasons. The primary reason is that too many organizations treat enterprise architecture as an engineering discipline expecting rigorous analysis and precise results. One should understand that an enterprise is an extraordinarily complex system of people, processes and technologies that interact in innumerable ways. We lack the science to accurately and substantively model these interactions. Perhaps a better label for traditional enterprise architecture (EA) is enterprise “art-itecture” as the architectural renderings help to communicate intent and direction.
In August, Gartner announced emergent architecture as a “new approach for enterprise architecture.” Along with their press release, it identified seven properties that differentiate emergent architecture, also referred to as middle-out EA and light EA from the traditional approach to EA:
The 7 Properties of Emergent Architecture according to Gartner:
1. “Non-deterministic: In the past, enterprise architects applied centralised decision-making to design outcomes. Using emergent architecture, they instead must decentralise decision-making to enable innovation.”
2. “Autonomous actors: Enterprise architects can no longer control all aspects of architecture as they once did. They must now recognise the broader business ecosystem and devolve control to constituents.”
3. “Rule-bound actors: Where in the past enterprise architects provided detailed design specifications for all aspects of the EA, they must now define a minimal set of rules and enable choice.”
4. “Goal-oriented actors: Previously, the only goals that mattered were the corporate goals but this has now shifted to each constituent acting in their own best interests.”
5. “Local Influences: Actors are influenced by local interactions and limited information. Feedback within their sphere of communication alters the behaviour of individuals. No individual actor has data about all of an emergent system. EA must increasingly coordinate.”
6. “Dynamic or Adaptive Systems: The system (the individual actors as well as the environment) changes over time. EA must design emergent systems sense and respond to changes in their environment.”
7. “Resource-Constrained Environment: An environment of abundance does not enable emergence; rather, the scarcity of resources drives emergence.”
Commentary re: these 7 Properties
1. Non-deterministic. The centralization vs. decentralization decision is one that has been around for quite some time and will remain one of the key choices that management will need to make and monitor. The main advantage of centralization is economies of scale; the main disadvantage is lack of customization which manifest in a variety of way including, the inability to innovate. In my client experience and research, I’ve seen traditional EA work well in centralized organizations as there is a often a business incentive to collaborate. The lack to collaborate in decentralized organizations leads to problems in realizing the value of EA. Hence, I’m skeptical of this property as explained by Gartner.
2. Autonomous actors. Successful enterprise architecture programs do not “control” but “influence” or “guide” decision making. Every individual in an organization, through her daily decisions affects the architecture of the enterprise. A well-seasoned enterprise architect provides a decision making framework and helps guide the organization in its decisions.
3. Rule-bound actors. As alluded to above, successfully enterprise architecture programs seek conformance but not necessarily compliance.
4. Goal-oriented actors. Each organization has a primary goal (e.g., increase company revenue 10% in 2010) supported by a hierarchy of sub-goals (retain > 95% of existing customers in 2010) that, in well managed firms, cascades throughout the fabric of the organization. Thus, the shift that Gartner mentions is really a move towards better alignment of corporate and individual goals, a trend that has existed for over a decade now but has been difficult to implement without enterprise employee performance measurement software.
5. Local influences. Certainly, it’s rare to find even an uber-enterprise architect who understands in detail each enterprise architecture layer (i.e., business, application, data, infrastructure, etc…) about an organization, let alone a group of individuals who are well-versed in all of these aspects. Certainly, it’s incumbent upon the enterprise architecture team to help coordinate, make sense of, and communicate the multitude of local conversations and decisions across the organization.
6. Dynamic or Adaptive Systems. The “sense and adapt” moniker is long in the tooth at this point. More progressive firms are targeting a “predict and impact” approach to business philology and thus system design. The powerful combination of service-oriented architecture and cloud-based computing helps in designing an enterprise that can quickly address many “avenues of evolution.”
7. Resource-Constrained Environment. The current economic environment leads to doing more with (even) less. However, I’m not certain that giving people more work do with less resources drives emergence. It’s more likely to drive poor enterprise architectural quality.
Summary Points:
§ Traditional enterprise architecture works well in highly-centralized organizations since collaborative decision-making is the modis operandi
§ Enterprise architects should adopt a conformance not compliance attitude – i.e., the “sheriff is in town” is not a good attitude to sport
§ Enterprise architecture models are abstractions of the real world. They are not precise by nature. However, they are useful in expressing intent as well as educational purposes.
Posted By : Ray Bordogna
Tags: cloud computing