Agile Scrum, XP, Kanban, Lean, DevOps and other cutting edge SDLC generally share the common goal of faster frame rate for delivery of functionality. “Deliver smaller, but complete, slices of functionality faster,” according to the core guiding principles, “and benefit from stakeholder feedback collected as early as possible.” Among other impacts, transitioning to these newer lifecycles puts tremendous pressure on the traditional practice of architecture.
Functional architecture, solution architecture, application architecture, integration architecture, and in particular, enterprise architecture, are traditionally areas where companies will invest a lot of time in planning. And few people will question why it is important to be cautious in charting a 5-10 year course for major systems. This work is at the heart of the value proposition of enterprise software development.
So, how can long-term planning and deep insight be accommodated in the much shorter delivery cycles recommended by the latest systems of development?
In fact the answer actually predates many of the new approaches: do just enough architecture in just enough detail, to meet the immediate need for development. And while doing so, make progress toward a coarsely defined ultimate vision for the future.
Or, as stated by Bente, Bombosch, and Lagande in Collaborative Enterprise Architecture (2012):
The most efficient EA [Enterprise Architecture]… exercises just so much control that the organization operates at the edge of chaos- structured enough not to let the IT slip into anarchy but not so rigid that it is locked into bureaucratic permafrost.
Agreed. Although I’m not sure the term edge of chaos will help create consensus! But in the end, strong architecture absolutely has a place in agile and other new SDLC. Getting agreement on this point is the first step.
Recent Comments