Supported configurations

Configuration	Streaming tables for Apache Iceberg
Cluster / broker type	Amazon MSK Provisioned with Express brokers only
Input format	JSON or JSON_SCHEMA_GSR
Output format	Apache Iceberg tables; Parquet files with ZSTD or Snappy compression
Schema source	AWS Glue Schema Registry (required)
Schema evolution	Not supported
Partitioning	Time-based (TIME_HOUR)
Data freshness	5–15 minutes (default 10)
Storage class	Managed by S3 Tables

Note

Input formats: JSON is plain JSON objects — you provide the Glue Schema Registry ARN that defines the schema. JSON_SCHEMA_GSR is GSR-serialized JSON, where the schema ID is embedded in each record.