Which compute workload can be migrated unchanged and remain performance-optimized on GCP?

Hadoop/Spark on Cloud Dataproc Regional with high availability is the correct answer because Cloud Dataproc is a managed service specifically designed for running Apache Hadoop and Spark workloads. It offers automatic scaling, high availability, and integrates well with other GCP services, ensuring performance optimization without requiring significant changes to the underlying applications. Web applications running on App Engine Standard would likely require code changes to fit the App Engine environment, as it's a platform-as-a-service (PaaS) with specific language runtimes and constraints. RabbitMQ on an unmanaged Compute Engine instance group would require manual management and scaling, which is not performance-optimized by default. Jenkins, monitoring, bastion, and security scanners on custom Compute Engine machine types would also require significant manual configuration, management, and optimization efforts, making them not "unchanged and performance-optimized" out-of-the-box.