An Agile Enterprise Architecture (EA) Delivers Critical Business and Mission Agility

While working with a recent partner, the question came up; “What changes are made to the EA approach if agile methods are required, or otherwise heavily encouraged?” The initial answer at the time was “Not many – we already have an agile approach to EA embedded in our Oracle Enterprise Architecture Development Process (OADP), and our Oracle Enterprise Architecture Framework (OEAF) is independent of project management and project development approaches.”

Our OADP has always been agile and therefore supportive of business and government agility – particularly in the current context of severely constrained budgeting cycles. We firmly believe in a “just enough, just in time” philosophy, with collaborative insight and contribution across teams and leadership, and delivery of EA artifacts or guidance tuned directly to prioritized results. This means strategic, useful and reusable guidance modeled and delivered in a manner that supports both longer-term initiatives and near-term objectives.

EA delivered as an agile approach, however, does require continual line-of-sight traceability back to the IT investment strategy – which in turn is aligned to the business strategy.  

In other words, a Sprint Iteration approach might be justified (i.e. using the “Scrum” strategy), from all relevant perspectives, to quickly establish a reusable process and metadata model for a common agency function – like “Document Routing and Approval” (DRA). The output might be required to inform a software solicitation (i.e. to explain the requirements).  The output might be to establish a reference model and basic governance (business rules) for identifying and improving process efficiencies around the agency where DRA is occurring.

The actual need for this EA artifact (or “Product”, in Agile terms) may be driven from an unanticipated mandate or regulatory change, and therefore require rapid response.  The need may also be limited in scope to only a portion of the agency’s business (i.e. those who actually know they need it).

So, an EA Sprint will work, and deliver what’s needed quickly and effectively to the target audience.  The highest return on investment (ROI) in this exercise, however, only exists if actual Enterprise traceability and impact assessment occurs. In other words, an agile EA output with a strategic Enterprise outcome.

Note this is a common misunderstanding for Agile software development; Agile programming and project management may deliver useful, rapid and cost-effective “features” from a Backlog of priorities, but much of the supporting infrastructure, integrations and organizational change isn’t delivered using Agile methods, but must evolve in a more strategic, methodic manner.  Preferably with EA guidance.

Here’s what should happen.  The common DRA process, metamodel and business rules begin to shape, in a somewhat parochial “requirements-driven” context, heavily leveraging the impacted SMEs for a short period of time. As this occurs, the Enterprise Architect and stakeholders begin mapping and comparing the DRA process design (at appropriately coarse levels of abstraction) to any similar that may exist within the agency, or among agency partners or stakeholders.  This may require some additional outreach and communication.  The EA may find additional SMEs, risk factors, standards, COTS DRA solution accelerators, overlapping data management projects, etc. – essentially other activities or resources that can be used or might be impacted.

The Enterprise Architect is the Scrum Master!

Strategic oversight and influence is therefore brought to bear on the EA sprint, and by leveraging EA methods, the impacts to the rest of the organization plus any modifications to the focus EA artifact can be addressed – entirely within standard and expected IT Governance. The EA artifact development is a Sprint, but actually leverages our lifecycle methodology – from Business Context through Current and Future States, and then Roadmap (i.e. Transitional Architecture) and Governance.  The EA Sprint may actually kick off or modify a more holistic EA maintenance process.

We are therefore avoiding an “agile everything” philosophy, though we’re delivering agile results.   We contribute over-arching guidance and process for both the DRA project and the organization as a whole, to make sure that all projects underway are still aligned to meet the needs of the business and IT investment constraints.

This is essentially what we believe in applying our EA process, over time or during more Agile response cycles; always raise and maintain focus on the business strategy and drivers to guide the investment of IT budget into those areas that affect the business most – or that are the most immediate priority, such as described above.

Thanks to Oracle Public Sector Enterprise Architect Director Bryan Miller for contributing to this article!

Enterprise Architecture – Death by Meta Models

Last week I absolutely astounded having read a project report for a follow on Enterprise Architecture engagement that a former colleague had undertaken after I had previously set up the Enterprise Framework and the strategic direction for engagement wi…

Enterprise Architecture – Death by Meta Models

Last week I absolutely astounded having read a project report for a follow on Enterprise Architecture engagement that a former colleague had undertaken after I had previously set up the Enterprise Framework and the strategic direction for engagement wi…

Categories Uncategorized

Danish metamodels

My Danish friends @gotze and @aojensen comment on the latest release of OIO EA, which is a national enterprise architecture framework and meta-model published by the Danish Government Agency for Digitization.

Both John and Anders feel that certain key artefacts have been placed at the wrong layer of abstraction. John writes

“In my view, Business Rules should not be located at the strategic level at all. I would argue that Business Rules primarily “belongs” to the Business sub-architecture domain.”

What is the basis for this argument? Anders points out a consequence

“business rules are located in the government strategy layer and thus tightly coupled to the long term vision of the government agency”

“Business rules are operationalisations of the long term strategy and strategic intent.
Whilst the vision, mission, and purpose of the enterprise do not change very often (i.e. provide the best available services our citizens), the business rules and processes involved in realising this will definitely change.”

and therefore

“Business rules belong in the business architecture.”

Thus Anders is basing his argument on a statement about the frequency of certain classes of change.

This statement appears to be empirically testable, although I know from my own experience that it is a lot difficult than one might think to gather data to test this kind of statement.

Part of the problem of measuring rates of change is that we don’t have a particularly robust theory of change in the first place. Let’s look at an example. From time to time, perhaps every year, Steve Ballmer restates the vision of Microsoft. Obviously he doesn’t use exactly the same words every year. And of course Microsoft-watchers will seek to interpret even the slightest change of wording or emphasis as a sign of a strategic change in direction. So even if Ballmer himself insists that the vision hasn’t changed, we might not believe him. Looking back in time, we might find that major changes in direction had already been hinted at in previous years. So at what point does an apparently minor change in wording become a substantially new vision?

Conversely, when a company has been exposed as unethical, the CEO will go public with an apology and an assertion of a new ethical vision. (Recent example: Barclays Bank.) We might not believe him either.

In both cases, we will probably judge whether there is a new vision or not by observing whether the company behaviour and rules changes or not. (And this is not just external observers – Microsoft and Barclays employees and managers are also making these judgements.) So the rate of change of vision might be epistemologically indistinguishable from the rate of change of behaviour.

However, despite the difficulties in conceptualizing and measuring change, I think it does make sense to derive architectural layers from the idea that certain things have a characteristic rate of change, and that things with a different rate of change should be in different layers. This means that there is at least a possibility of subjecting an architecture to empirical evaluation. I have published this idea in articles for the CBDI Forum, and suggested that architectural theory needs to be based on the Pace Layering principle

In contrast, Anders’ appeal to the IAF seems to be purely an argument from authority. The IAF establishes some “fundamental” categories, and so any framework that deviates from these categories must be wrong. I think this line of argumentation is weaker. Even though you may assert some attractive consequences of following IAF, I cannot see any reason for believing that these consequences follow only from IAF and not from any rival framework.

Frameworks and categories may be embedded in metamodels. But how do we know what is the basis for choosing between alternative metamodels?

John Gøtze, Metamodels (January 2013)
Anders Ø. Jensen, Enterprise Architecture and abstraction layers (February 2013)

Ethics, Barclays and totalitarianism (Catholic Commentary January 2013)
Barclays boss tells staff ‘sign up to ethics or leave’ (BBC News January 2013)
Did Barclays suffer an Ethics Meltdown? (CSR Zone, July 2012)
Sure Kamhunga, Barclays to re-examine its core values(October 2012)
Naven Johal, Barclay’s Does Something Right! (January 2013)

Updated 29 April 2013

Attack of the Vegan Zombie Business Cases

The Business Analyst and the Project Manager have yet another impossible timeframe forced upon them, which they accept.  Faced with the prospect of losing the confidence of their Project Sponsor, they somehow manage to squeeze out their Business Case.  The Business Case is such a success that they circulate it to the usual suspects for […]

Managing Business Transformation

Just putting together the material for my new workshop next week.

This is the third day of my Business Architecture series. The first two days cover the six business architecture viewpoints. The idea is that people can tale these separately or together.

Day One – Modelling Business Operations 
Exploring process quality issues using the Activity Viewpoint, Knowledge (Information) Viewpoint and Motivation (Purpose) Viewpoint.

Day Two – Modelling Business Organization 
Exploring business relationships and strategy, using the Capability Viewpoint, Responsibility (Organizational) Viewpoint and Cybernetic Viewpoint.

Day Three – Managing Business Transformation 
Process guidelines and roadmap for business architects to analyse and manage structural change in large complex organizations.

Read more »

Metamodels

The Danish Agency for Digitisation has announced some coming updates of the national enterprise architecture framework and reference models. In a consultation draft about these, Et fælles overblik, the agency also introduces the OIO EA metamodel. The consultation also involves an update to STORM, the Service and Technology Reference Model. All documents are in Danish. Interested parties can submit comments to the agency until 14 …read more