Comparing The Aws Lambda Serverless Function Deployment Models

In recent years, the field of serverless computing services has experienced significant growth. Currently, the majority of cloud service providers offer serverless solutions, with Amazon Web Services (AWS) providing Lambda as their Function as a Service (FaaS) platform. Two primary implementation methods for application (function) delivery exist for AWS Lambda: the compression of source code and associated requirements into a ZIP folder, or the utilization of a container image encompassing the application and its dependencies. The chosen deployment approach can significantly impact function performance, cost-effectiveness, and initialization time. This work examines these critical metrics to compare the aforementioned deployment methodologies, with the objective of determining the most suitable approach. Empirical research conducted on AWS Lambda demonstrates that functions deployed via compressed ZIP folders exhibit superior characteristics, particularly in terms of initialization time during cold-start scenarios and overall cost-efficiency.