A data analysis program is running on a Spark cluster of 5 nodes. The data is partitioned on all 5 nodes. For each of the observations below. suggest what operations can the programmer perform to optimise the performance ? Name the operations in each case and describe in brief what they achieve [Marks: 6] 1. All 5 nodes are not always used. Some data/RDDs may use 4-way partitions. 2. Some of the operations could be faster because they repeatedly access same data. 3. Some of the data is used only once but is contributing to high memory usage