You are a machine learning specialist tasked with deploying a real-time inference model using Amazon SageMaker. The model must maintain performance while handling variable load during peak and off-peak hours. Which deployment configuration should you use to manage these fluctuations efficiently?