Modernizing Enterprise Architecture for Outcomes

Rohrer's presentation aimed to redefine the field, adapting it to modern demands and drawing inspiration from Dave Farley's principles of modern software engineering to focus on delivering outcomes.

The framework begins with 'A' for aligning people, value, and technology, emphasizing their interconnectedness.

'B' stands for better value, sooner, safer, happier, which refers to the desired outcomes of enterprise architecture, while 'C' highlights the importance of continuous and conversational interactions.

Finally, 'D' represents DevOps at enterprise scale, and an implied 'E' refers to evolutionary enterprise architecture, underscoring the need for adaptive systems.

The traditional software development model involved lengthy planning phases, often spanning months, before any code was written.

After coding, the product was often 'thrown over the wall' to a test team, leading to siloed handovers and prolonged delivery cycles.

Modern approaches like Agile and DevOps have transformed this, enabling organizations to move from an initial idea to production in a matter of hours or days, signifying a critical shift from 'project mode' to a continuous 'product mode' where systems are constantly evolved and refined.

In 2008, Roger Sessions warned that unmanaged systems could evolve into a 'spider's web,' leading to severe IT complexity and significant costs.

This challenge was exacerbated by the mistaken adoption of civil engineering metaphors for software delivery in the 1990s, which failed to account for software's inherent flexibility and constant change.

Complexity continues to be a major obstacle for modern organizations, particularly those managing multiple diverse teams, making effective enterprise architecture crucial for mitigation.

Alan Kelly further observed that existing software architecture can restrict attempts to reorganize an enterprise, creating a feedback loop where initial designs influence future structures.

Rohrer explained this dynamic using Winston Churchill's analogy of shaping buildings that, in turn, shape us, underscoring the need for modern enterprise architecture to intentionally design both the organization and the system as deeply interrelated components.

Business value streams, organizational structure, and IT architecture must align to foster sustainable work flows, according to the principles of enterprise architecture.

This alignment must extend across all levels of the enterprise, maintaining a fractal, self-similar consistency.

Drawing on concepts like Team Topologies, the approach advocates for autonomous teams that own specific value, often conceptualized as 'two-pizza' or 'full-stack' teams, which aligns with Dan North's idea of 'software that fits in your head' to empower business capabilities.

Instead, the focus should be on creating "team-sized software" that development teams fully own, deploy, and test independently.

In this model, architects function more as coaches within teams, guiding decision-making rather than dictating specific technical solutions.

Mel Conway's observation highlights that systems inevitably mirror an organization's internal communication structures, suggesting that the initial design is rarely optimal and often requires subsequent changes.

This necessitates a flexible and lean organizational structure that can adapt and evolve systems over time, as effective architecture requires the simultaneous alignment of people, technology, and value.

The ability to iterate and refine system concepts is crucial for achieving optimal design and maintaining relevance in a dynamic environment.

Enterprise architecture must move beyond traditional focuses such as standardization, cost reduction, or the adoption of specific technologies like Kubernetes or AI.

Rohrer emphasized that mere outputs, such as migrating to the cloud, do not equate to tangible business outcomes.

Instead, the goal should be a balanced scorecard approach, aiming for "better value, sooner, safer, and happier" outcomes for all stakeholders, including customers, colleagues, and the climate.

These teams, encompassing front-end, back-end, and database administration expertise, enable them to own an entire valuable business capability from concept to deployment.

This integrated approach significantly reduces coordination costs and internal dependencies, as eliminating just one dependency can reduce risk by 8x, decrease cost by 5x, and improve efficiency by 4x.

Adding complexity to an architectural system often appears inexpensive at the outset, but reversing or removing that complexity is an asymmetric and far more challenging task.

This difficulty is particularly pronounced in distributed systems, where changes ripple across numerous components, making the strategic imperative to avoid introducing unnecessary complexity from the start.

Rohrer advocates for treating a request for a new service as a conversation initiated via a pull request, moving away from traditional paper-based reviews.

The architecture team assesses the impact of such changes on overall system complexity and service-to-service connections.

This pull-request-based governance also applies to exceptions, such as preferring synchronous architecture over standard asynchronous patterns, ensuring that governance focuses on reviewing actual system modifications rather than merely rubber-stamping proposals.

This approach fosters continuous conversational governance, replacing rigid, one-off, phase-gated reviews with ongoing dialogue.

The model is designed to scale by maintaining a lean core architecture team that is supported by a broader community of practitioners, reducing the need for manual approvals and accelerating development.

Modern architects must bridge the gap between design and site reliability engineering (SRE) or service management silos, collaborating closely with these teams.

Ruth Malan's framework of 'intention and reflection' highlights the iterative use of diagrams to both design and observe systems, emphasizing the need for architects to move beyond predictive models.

Architects now utilize observability tools to learn from actual system behavior in production, acknowledging that purely predictive architecture is insufficient for systems designed to operate for decades.

In complex environments, product teams must balance both technical and product ownership, necessitating rapid feedback loops to address challenges effectively.

The 'probe-sense-respond' cycle is crucial for solving complex problems, involving continuous wiring, development, operation, building, and running of systems.

This emphasizes a shift from traditional project-based work to a product-centric approach within modern enterprise architecture, ensuring agility and responsiveness at enterprise scale.

Evolutionary architecture, drawing on the concept of punctuated gradualism, acknowledges that system evolution occurs through periods of steady, gradual change punctuated by significant, strategic shifts.

While fitness functions are effective for microservices, they pose challenges in highly distributed systems.

To combat complexity debt, continuous funding, typically around 20% of the budget, is essential.

The 'strangler fig' pattern is favored over complete system rewrites to avoid the 'second system syndrome' and the substantial financial and operational risks associated with full rewrites, which are rarely viable without significant spare cash.

The Standish report reinforces this by advocating a shift from a project-based mindset to one focused on continuous implementation, which aligns with the dynamic nature of competitive markets.

The focus should be on achieving 'better value sooner, safer, happier' outcomes, rather than traditional enterprise architecture objectives. Furthermore, conversational governance is crucial, replacing rigid, phase-based approaches for more agile and continuous improvement.