chapter ten

10 When hexagons talk: Integration modes and the external architecture

 

This chapter covers

  • Distinguishing inbound and outbound integration
  • Placing an external system as a driving or a driven actor
  • Telling a request for work apart from a fact that something happened
  • The three integration modes: synchronous calls, events, and orchestration
  • Telling a business process apart from a technical process
  • Choosing a mode from the interaction's consistency and time needs
  • Keeping ownership clear across the boundary
  • Mapping the external architecture as contracts for WriteFlow

WriteFlow no longer runs alone. A hypothetical partner Content Management System (CMS) asks it to reopen an article, a publishing platform waits to hear when one is approved, a newsletter and an analytics service react to that same fact, and a workflow engine runs the editorial pipeline that now calls back into WriteFlow. Part 1 established this boundary to keep infrastructure concerns out of the domain core. In this chapter, the boundary also makes changes in interaction modes, transactional consistency, and data representation visible in one identifiable part of the architecture.

We will follow the CMS’s request to reopen an article through the chapter’s integration decisions and onto WriteFlow’s contract map. The approval event and editorial pipeline are separate interactions that illustrate other integration choices.

10.1 Two directions at one boundary

10.2 What crosses the boundary: Requests and facts

10.3 The three integration modes

10.3.1 Synchronous calls

10.3.2 Event-driven integration

10.3.3 Orchestrated integration

10.4 Choosing a mode

10.5 The rules that survive every mode

10.6 An example of external architecture for WriteFlow

10.7 Summary