When is Hadoop better than PySpark?

Hadoop excels over PySpark for massive, batch-oriented tasks that don't require in-memory processing or low latency. Think large-scale ETL or data archiving where disk-based processing is sufficient and cost-effective. It's also a strong choice when your existing infrastructure heavily relies on the Hadoop ecosystem and you need a robust, fault-tolerant system that manages resources efficiently without the memory overhead inherent in Spark.