You are deploying a machine learning model for a financial forecasting application that requires batch processing of large volumes of data at the end of each quarter. The deployment must efficiently handle these sporadic, compute-intensive workloads. Which configuration should you choose for the batch deployment?