On how to realize fault tolerance in Hadoop/spark

Hadoop uses data replication for fault tolerance (I/O high)

Spark uses the RDD data storage model to achieve fault tolerance.

The RDD is a collection of read-only, partitioned records. if a partition of an RDD is missing, the RDD contains information about how to reconstruct the partition. This avoids the need to use data replication to ensure fault tolerance , thereby reducing access to the disk. With Rdd, the next steps do not have to be recalculated or loaded from disk if the same data set is required.

On how to realize fault tolerance in Hadoop/spark