Saturday, August 30, 2014

Can Enterprise Architecture be Agile?

IT practice changes quite a bit and certain practices come and go. About 10 years ago, Enterprise Architecture (EA) surged to the forefront of the IT community primarily because it was adopted as part of the key acquisition processes connected to the Federal Government. The law that brought this about was called the Clinger-Cohen Act and it provided portfolio management expectations for IT which were designed to help avoid large-scale IT project failures (ones costing the taxpayers more than $100 million) but also applied to all IT investments. Not too long after this law was passed in 1996, the Department of Defense and other agencies redesigned their IT processes / methodologies to become compliant with Clinger-Cohen. Soon after that the practice of EA became very much in demand.

EA was around before this of course, namely through the introduction of the Zachman framework in the late 80's, early 90's, but most organizations hadn't adopted it yet. The federal mandate provided an excellent incentive both for federal and commercial adoption. It also created some issues however, namely that much of the early work in EA became tied to a more procedurally intense form of practice - in other words, to many it seemed to be a bit bureaucratic in nature. The perception quickly became that EA was not an easy undertaking or something that could be used to produce quick results. This is similar to what happened to E-learning at roughly the same time for about the same reasons - government adoption and stewardship was both a blessing and curse to the industry.

So the following question is now often posed within IT in regards to Enterprise Architecture - is it possible for EA to be Agile? In most quarters, the immediate reaction would be a resounding - no. But perhaps maybe we're being too harsh in our initial judgement. First, though, we need to define what Agile would mean in an EA context:
  1. It would require rapid turnarounds for designs and decisions.
  2. It would require the ability to integrate both with traditional (waterfall) methodology and Agile development methodology.
  3. It would need to support and promote collaborative problem-solving within the organization using it. In other words, an EA group doesn't become another island or silo, it is the mechanism that bridges all of the silos.
Based on these definitions, EA can and definitely is Agile if practiced correctly. The following articles provide examples of how various aspects of EA can be integrated within an organization and remain Agile:


Many of the articles included in this blog were produced from projects that followed an Agile EA approach, including the Intelligent Healthcare framework, Semantic COP and Governance as a Service. I've applied Agile EA in the following domains thusfar:
  • Defense
  • Aerospace 
  • Healthcare
  • Retail
  • Manufacturing 
  • Finance
I've also applied to the following IT focus areas as well:
  • Cyber-security
  • Data Management
  • Services / application design
  • Portal / CMS
So returning to the original question - can EA be Agile - my answer is yes. The key to making that happen involves the following assumptions:
  1. Clearly defined expectations for the both what the EA group or architect must accomplish.
  2. A willingness to improvise rather than following industry patterns or methodologies too closely.
  3. The ability to work to deadline - to engineer a design around the constraints.
  4. A willingness by the organization to allow the silos to actually work with one another (with an architect as mediator) - this is far less common than you might imagine.
  5. The ability to develop architecture that is easily translatable and traceable to implementation level designs. In other words, alignment of all design related efforts must be built into the Agile approach.
  6. The ability to make decisions quickly. 
  7. A willingness to experiment. Many architects and organizations make the mistake of separating EA into a logical or somewhat abstract exercise and not allowing it to be involved in actual prototypes or POCs. The best way to ensure that key design decisions make sense is through an understanding of the technology in question.

Copyright 2014,  Stephen Lahanas




#ITarchitectureJournal
#StephenLahanas
#Semantech-Inc


0 comments :

Post a Comment