In the data space there is this new movement of ZeroETL (started by AWS) that advocates the replacement of traditional data movement pipelines (generally implemented using Spark and Airflow) with real-time CDC replication. What is your take on it ? Did anyone fully implemented it ?