Apache Spark Unpersist and Persist

I have been running a very large use cases and due to resource constraint, I have to limit the no. of data being processed at a given time. So running a counter for loop such as 1 to 10 and for every run I process, say 200000 rows.

My question is, if my source has 1M rows, I would end up processing for 5 runs (1M/ 200000). And I have multiple sources also. So, for eg.

Source A has 1M rows
Source B has 2M rows.

When i am in the loop for each Source, say,

for (source in Sources) {
  val rddSource = spark.read.table(source)
  rddSource.persist
  rddSource.count // to persist the above rdd

  while (counter < 5) {
       //process every 200000 rows with the above persisted rddSource

    }
}

I want to persist the data as shown above.that will help to run Source A for every 5 time each with 200000 rows.

But for next iteration of for loop with the Source B, does it replace the previously cached {rddSource} or should I need to have something like this

for (source in Sources) {
      val rddSource = spark.read.table(source)
      rddSource.unPersist
      rddSource.persist

      rddSource.count // to persist the above rdd

      while (counter < 5) {
           //process every 200000 rows with the above persisted rddSource

        }
    }