In Apache Spark, checkpoints are used to allow the program to run all around the clock. It also helps to make it resilient towards failure irrespective of application logic.