Software Engineering
System Integration Patterns for Organisations With Legacy Systems
Replacing core systems is rarely the first move. Integration patterns that unlock new capability while existing platforms stay in place.
Wrap before you replace
Placing a well-documented API in front of a legacy system lets new applications consume its data safely without waiting for a rewrite. It also creates a seam that makes eventual replacement far less disruptive.
Choose synchronous or event-driven deliberately
Request-response integration suits operations that need an immediate answer. Event-driven messaging suits workflows where systems must stay eventually consistent without being coupled to each other's availability.
Mixing the two without a clear rule produces integrations that are hard to debug and harder to change.
Make data ownership explicit
Every entity should have one system of record. Integration problems that look technical are frequently governance problems in disguise, where two platforms both believe they own the same data.
