Why another big-bang replacement can recreate the same problem

Why another big-bang replacement can recreate the same problem

When architecture knowledge becomes difficult to use, replacing the repository can feel like the decisive move. A new platform promises a cleaner model, better search and a fresh start. The programme is visible, fundable and easy to describe.

But the repository is rarely the whole problem.

The knowledge needed for change also lives in documents, operational systems, integration catalogues, local models and the experience of people who understand why the landscape looks the way it does. Moving one collection of models into a new tool does not reconnect that context. It may simply move the centre of fragmentation.

As application landscapes become more composable, that matters even more: the context needed for change increasingly exists across systems rather than inside one of them.

Migration is not the same as modernization

A migration answers a technical question: how do we move information from one environment to another? Modernization asks a more demanding question: how will this knowledge become more useful for the decisions the organization must make?

Those questions lead to different designs. A platform-led migration often begins with the destination environment and what can be moved into it. Modernization begins one step earlier: with the decisions, relationships and evidence that must survive the transition.

This matters because architecture information is not valuable merely because it has been copied. Its value depends on structure, provenance, access rules and the ability to relate it to other sources. If those conditions disappear, a technically successful migration can still reduce the organization’s capacity to reason about change.

The same risk appears when a programme attempts to model everything again from a blank sheet. The new environment may be elegant, but it begins with less organizational memory than the environment it replaces. Teams then reconstruct context through workshops and manual discovery while delivery continues around them.

Start by releasing knowledge from the platform

A more progressive route is to separate the useful knowledge from the product that currently contains it.

In one repository-preservation situation, the useful outcome was not to keep a retired platform running. A one-time export became a governed source. Objects, relationships and diagrams could then be inspected through a new graph experience, searched and reported on without pretending that the old repository was still operational.

That distinction is important. Preservation does not mean freezing the past. It means retaining enough structure and context for the past to remain intelligible while the organization moves forward.

The same principle applies to a living architecture environment. Before selecting a replacement, identify what people actually need to access, which relationships must remain trustworthy and which decisions depend on them. Some information may move. Some may remain in place and be connected. Some may no longer deserve the cost of migration.

Modernize progressively, not indefinitely

Progressive modernization is not an argument for postponing difficult choices. It is a way to make each choice with better evidence and a smaller blast radius.

An organization can begin by making one important knowledge source accessible. It can connect that source to the operational information needed for a specific decision. It can then assess whether the new capability should be retained, expanded or replaced. Each step produces a usable result rather than waiting for the completion of a multi-year platform programme.

This is also a more honest way to manage architecture maturity. Run, Access, Connect and Insight are not stages that every organization must purchase in sequence. They are possible entry points based on the environment and the next decision. What matters is that each step increases usable context rather than creating another isolated destination.

The alternative to a big bang is not a collection of disconnected pilots. It is a controlled path from the architecture environment that already exists to one that supports better decisions.

A new platform may be part of that path. It should not be mistaken for the path itself.