You are planning to migrate numerous Apache Spark jobs from an on-premises Apache Hadoop cluster to Google Cloud, aiming to use managed services for job execution to avoid managing a persistent Hadoop cluster. With a tight deadline and the objective to keep code changes to a minimum, what is the most efficient approach to achieve this migration?