Describe a specific instance when you had to scale a service in response to an unexpected traffic surge. What considerations did you make?