Landscapes

Two maps, and the value is entirely in the difference between them.

The context map is intent: how the bounded contexts are meant to relate, in the vocabulary of the strategic patterns. Who is upstream, who conforms, where an anticorruption layer is doing real work.

The system map is reality: which running systems actually talk to which, over what, observed rather than asserted.

Conformance is the comparison, and it is why both exist.

Why two, and not one

Each map on its own has a characteristic failure.

A context map alone is a wish. Nothing contradicts it, so it survives indefinitely, gaining authority as it ages. Everyone has seen the four-year-old diagram that no longer describes anything and is still on the wall.

A system map alone is a network diagram. Every arrow is true and none of them is judged. It can tell you that claims-platform reads from legacy-policy-master; it cannot tell you that this arrow should not exist, because it has no notion of should.

Drawn against the same catalog, the two produce a third artefact for free: the list of places where the estate has quietly stopped implementing the business.

The four verdicts

Every edge, in either map, gets one:

Matches — an observed integration realises a declared relationship, in the way the pattern says it should.

Drifts — it realises it badly. The relationship exists, and its implementation violates the pattern: an anticorruption layer that was never built, an adapter that gets bypassed, a shared kernel implemented as an overnight fuzzy match.

Unmapped — a real, observed integration with no relationship behind it. Two systems are talking and nobody declared that their contexts should. This is always a modelling question rather than an integration one: either the relationship is legitimate and undeclared, or the boundary is in the wrong place.

Unobserved — a declared relationship with nothing observed behind it. Either it was agreed and never built, or the collector cannot see it. Those have different owners, and the map cannot tell them apart on its own.

Findings are business statements

The discipline that makes this section useful rather than another dashboard: a finding is written as a sentence about the business.

Not: claims-platform performs direct reads against the legacy-policy-master schema.

But: a claim can be assessed against cover that was added to the policy after the loss happened.

Both describe the same fact. Only one of them can be taken to somebody who can authorise the fix, and only one of them makes it obvious why it should be near the top of the list. The seed landscape’s worst finding is exactly this one.

What is real today

The landscapes page is populated with a worked example. The relationships are the ones a modelling exercise would produce; the integrations are what a collector would observe, written by hand.

The landscape collector that would gather them continuously, and the DDD mapper that would compute the verdicts, are designed and not built. Doing the comparison by hand once is still worth more than not doing it — it is how you find out whether the automated version would pay for itself.

In this section

• Context mapping — drawing the map of intent, and what belongs on it.
• System mapping — gathering evidence about what actually runs, and the blind spots of each source.
• Conformance and drift — judging the difference, and prioritising by cost.