Exploring journal-based CDC for InterSystems IRIS
What changes when data capture starts at the transaction journal rather than at the application layer?
InterSystems IRIS often sits at the centre of operational workflows, while downstream systems need updates in near real time. The integration challenge is not simply moving data: it is capturing meaningful changes without introducing excessive application coupling or database load.
Why investigate the journal?
Application hooks, scheduled exports and polling approaches can be effective, but they also introduce design trade-offs. A journal-oriented approach explores a lower-level source of changes that may reduce the need to instrument each application path.
The difficult parts
Decoding physical journal changes is not equivalent to publishing clean business events. A practical pipeline must handle transaction boundaries, record formats, data mappings, recovery, replay, security and downstream compatibility. Exactly-once effects require careful end-to-end design rather than a label on the transport.
Our direction
We are prototyping a CDC pipeline that interprets relevant IRIS journal changes and publishes structured change events to Kafka. The architecture is exploratory; publicly documented benchmarks, compatibility details and production guarantees will be provided only when validated.