Skip to content

Tools

One tool for the wall the model is built on, and one for checking whether anyone still respects it.

The first exists. The Event Stormer runs event storming and process modelling on the same board — big picture for discovery, process modelling for the commands and policies, software design for the aggregates underneath — because they are three depths of one notation rather than three exercises. The second does not exist yet: a mapper that holds the catalog and the observed estate at the same time and computes the difference, so its interesting output is never the map but the list of edges the map got wrong. Everything here serves one of those two jobs.

One of these is deployed. The other four are not.The Event Stormer is real and usable now — it is doc-hub’s board, and it runs two of this hub’s practices rather than one: event storming and process modelling are two of its three levels. For the rest, the designs are written and the interfaces are decided; the software is not built. The addresses those panels point at are configuration, so the day something is deployed behind one it takes a value change and a restart rather than a rebuild. Until then, thelandscapespage does by hand what the mapper would do continuously.

Tools

What goes in, what comes out

A tool with no named input is a drawing program, and a tool with no named consumer is a hobby. Both columns are filled in for every entry below, which is a cheap way to notice when one of these should not be built.

ToolDoes what nothing else doesReadsEmits
Event StormerRun both workshops — event storming and process modelling — on one wall, at whichever of the three levels the room is working at.A room, a facilitator, the people who do the work, and a declared level.An `.eventstorm` file: candidate bounded contexts from the big picture, the commands, policies and read models of a process, the aggregates under it, and every hotspot nobody could settle.
DDD mapperHold the catalog and the two landscapes in one place, and compute the difference between them.The catalog, the declared relationships, and the landscape collector’s observations.A context map, a system map, and a conformance table with an owner against every finding.
Landscape collectorReplace assertion with evidence about what the estate actually does.Service meshes, database audit logs, broker topic metadata, schema registries.Observed integrations with evidence, consumed by the DDD mapper.
Language workbenchKeep the ubiquitous language honest across contexts.Catalog language sets, requirement documents, API schemas.A per-context glossary and a collision report.
Storm reconcilerPlace a finished wall against the catalog, so its boundary crossings are computed rather than remembered.An `.eventstorm` file, and the catalog to place it against.Boundary crossings named with their compensating actions asked for, and new `context` cards raised as catalog candidates.

And the part a model can do

Some of this work is judgement and some of it is reading. Distilling a first pass at the subdomains out of documents the business already wrote, spotting that one word carries two meanings across four systems, or noticing that an integration nobody declared has appeared — none of those need an analyst, and all of them currently consume one.