At present dagger only support Apache Kafka as an input source. However, to analyze historical data relying on only Kafka is cumbersome. To facilitate easier and faster data analysis on historical data, we want to add support for reading parquet data as an alternate source