In the context of optimizing data distribution across nodes in a Microsoft Azure Synapse Analytics environment using a custom partitioner for a Spark RDD, which of the following factors is considered the least important?