In the context of Apache Spark, data skewness is a critical issue that can significantly impact the performance of data processing tasks. Consider a scenario where you are working with a large dataset that exhibits high skewness in the distribution of keys, leading to some partitions being heavily loaded while others are underutilized. This imbalance causes certain tasks to take much longer to complete, creating bottlenecks in your workflow. You are tasked with selecting the most effective technique to mitigate this skewness, considering factors such as cost, implementation complexity, and the potential for improving overall job performance. Which of the following techniques would you choose to address data skewness in this scenario, and why? Choose the best option from the following: