In Spark, if any partition of an RDD is lost due to the failure of a worker node, that partition can be re-computed using the lineage of operations from the original fault-tolerant dataset.

BY Best Interview Question ON 10 Jun 2020