You are at a large logistics and shipment company, hosting a shipment tracking application on Compute Engine VMs in the us-central1-a zone. To ensure the application remains operational during a zonal failure on GCP with minimal costs, what should you do?