Streaming data from Apache Kafka into Delta Lake is an integral part of Scribd’s data platform, but has been challenging to manage and scale. We use Spark Structured Streaming jobs to read data from Kafka topics and write that data into Delta Lake tables. This approach gets the job done but in production our experience has convinced us that a different approach is necessary to efficiently bring data from Kafka to Delta Lake. To serve this need, we created kafka-delta-ingest.
