Scaling for Peak API Consumption: Reference Architecture