Why point-to-point integration stops scaling
LeanSignal technical brief · Enterprise data & systems architecture.
Point-to-point integration is often the fastest way to connect two systems. It becomes expensive when the organisation repeats the same local optimisation across dozens or hundreds of systems.
The problem is the dependency graph.
Each additional connection carries mappings, policies, versions and operational assumptions. Those assumptions become dependencies for future change.
Shared contracts reduce repeated decisions
Federated architecture does not remove integration. It moves common semantics, metadata and policy into reusable contracts so integrations stop recreating the same decisions.
Architecture principle
Use point connections behind stable domain contracts rather than making each connection itself the enterprise architecture.
Signals that point integration has reached its limit
- Integration projects require the same mapping work repeatedly.
- Changes need coordination across unrelated application teams.
- Multiple “authoritative” copies of the same information exist.
- Replacing a vendor creates enterprise-wide migration work.
- Failure analysis depends on people remembering undocumented relationships.
General architecture guidance. Specific environments require adaptation to their security, regulatory, operational and technology constraints.