Scenario: You have an upstream process that writes data to Cloud Storage, which is later read by Apache Spark jobs running on Dataproc in the us-central1 region. The data could be stored anywhere in the United States. You need a disaster recovery plan that ensures a maximum of 15 minutes of data loss (RPO = 15 mins) and minimal latency when reading data. Question: What approach should you use to design a recovery process that meets the RPO requirement and minimizes latency?