You are tasked with designing a Dataflow pipeline for a batch processing job in a cloud environment. It's critical to ensure the continuity of the job even in the face of potential failures in multiple zones at the time of job submission. How would you approach this to enhance the resilience of your Dataflow pipeline against zonal failures?