How would you optimize cloud resource utilization for auto-scaling workloads with highly variable traffic patterns?
