In Confluent Cloud, navigate to the Connectors section and select a managed sink connector (e.g., Snowflake Sink, BigQuery Sink, or S3 Sink).
Provide the target system credentials, destination database/dataset/bucket, and the source Kafka topic(s) to consume.
Configure the input data format (Avro, JSON Schema, Protobuf) to match what the upstream ksqlDB query or Flink job produces.
For ksqlDB-originated topics, ensure the topic's value format is schema-registry-backed (Avro or Protobuf) so the connector can resolve schemas automatically.
For Flink-originated topics on Confluent Cloud, configure the Flink sink table connector properties to match the managed connector's expected topic format.
Monitor connector lag, throughput, and error metrics in the Confluent Cloud UI or via the Metrics API.
Known gotchas
Confluent managed connectors run on a shared or dedicated Connect cluster; their configuration properties and available connector versions may differ from self-hosted Kafka Connect — check the Confluent Cloud connector documentation rather than generic Kafka Connect docs.
Schema evolution must be managed through Schema Registry compatibility settings; a backward-incompatible schema change will break the sink connector until the connector configuration is updated.
Some warehouse sink connectors buffer records and flush on a schedule or size threshold; end-to-end latency from stream record to warehouse row can be minutes — do not use these for near-real-time query SLAs without understanding the flush interval.
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp