IIS IDR Integration via KAFKA

DataStage:
Kafka Connector in Information Server 11.7 does not support the use of schema registry or Avro JSON serialisation/deserialisation. IIDR (CDC) for Kafka on the other hand, produces Avro JSON serialised records into topics as default. Using Avro and schema registry have advantages for high volume and near-real-time integration use cases. At present, those benefits cannot be realised using IIS 11.7 Kafka Connector. More importantly, it does not support the default behaviour of IIDR for Kafka. This disadvantages the solutions and DataStage against competing technologies such as Talend which offers much superior support for Apache Kafka.

On other main issue using Kafka Connector in IIS 11.7 to consume data produced by IIDR for Kafka is the inability of Kafka Connector to handling NULL records. For physical deletes in a source, IIDR produces a NULL record at which point the DataStage job aborts.


CDC for Kafka:
On the IIDR (CDC) for Kafka side, there should be some flexibility for developer to select a output format and serialisation method without forcing down the Avro JSON path. For example, being able to produce string serialised or byte array serialised CSV records into topics.
Avro JSON requires a schema registry and it is not a generic Kafka SerDes (Serializer/Deserialiser). Although IBM has provided some user exits (KCOP - Kafka Customer Operation Processors) as a way around to this and produce records without a schema registry, using KCOP adds a processing overhead, since the producer first generates the record in Avro and then converts back by the KCOP to the desired format. In addition, only one KCOP can be used in a subscription and developer need to write custom KCOPs to combine any two or more non-standard CDC functions.