Assessing the Migration from FaaS to IaaS: Cost, Performance, and Challenges in AWS

Abstract

In cloud-native environments, the choice of service model is a critical factor in optimizing both operational outcomes and associated costs. This study examines the migration from a serverless Function-as-a-Service (FaaS) model— specifically AWS Lambda—to a monolithic solution deployed on Infrastructureas-a-Service (IaaS), more specifically Amazon EC2. Through a series of tests covering different types of applications and workload patterns, the performance and operational costs of both architectures are evaluated. The results indicate that while the FaaS model offers simplicity and automatic scaling, it may incur significantly higher costs in high-demand scenarios. In contrast, the IaaS model provides greater control and cost optimization potential, though it introduces additional challenges related to infrastructure management and maintenance. Moreover, the study identifies and characterizes the main technical and operational challenges associated with the migration process, outlining common patterns and best practices to guide future transitions between cloud service models.