OPTIMIZING AWS LAMBDA COLD STARTS THROUGH PRIMING: A TECHNICAL EXPLORATION

Abstract

This technical paper explores how Priming can be a deliberate tool to reduce cold start[1] time in AWS Lambda serverless services. Cold starts, indicating initial delays when executing a function after inactivity, often pose challenges in serverless systems. Using a heavy-weight dependency injection design or framework in Java systems exacerbates this difficulty. The article analyzes situations where desired response times are not attained, uncovering optimization insights even when devising concurrency models such as provisioned concurrency and Snapstart, which aim to reduce cold starts, are adopted. This article investigates the adoption of priming methods to decrease cold start times, thereby improving the overall performance of serverless apps. The author, coming from an AWS and Java background, is committed to using these technologies to express the concept throughout.