Comment by jamesblonde

DuckDB's sweet spot is for 'smallish' lakehouses. So, ingestion should not be via duckdb, but rather something like Python/DLTHub for small scale or Spark for large scale or Kafka/Debezium/Flink for streaming data.