Designing a Dataflow batch job: you want to mitigate multi-zonal failures at job submission time. What should you do?

Specifying a worker region using the --region flag ensures that Dataflow can automatically distribute workers across multiple zones within that region. This provides resilience against single-zonal failures during job execution. If one zone experiences an outage, Dataflow can continue processing in other available zones within the specified region, mitigating the impact. Submitting duplicate pipelines in different zones (--zone flag) is less efficient and doesn't provide automatic failover; you'd have two independent jobs. Setting the staging location as a regional Cloud Storage bucket is good practice for data durability but doesn't directly control the compute resource distribution for the Dataflow job itself. Creating an Eventarc trigger for resubmission is a reactive measure for job failure, not a proactive one for mitigating multi-zonal failures at submission time.