How would you choose between Kafka, Flink, Spark, Cassandra, and ClickHouse for a large-scale data system at Datadog?
