Empirical Evaluation of Cold Start Latency in Serverless Platforms
Vinay Raj, Himanshu Chhaparwal, Satwik Saale · 2024
Serverless computing has gained significant popularity as a cloud-based alternative for creating, testing, and deploying applications as serverless operations. Serverless computing has gained significant attention in both industry and academics due to the introduction of serverless platforms by major IT companies. The primary characteristics of this system include autonomous design, seamless deployment, and automatic scalability, with Function-as-a-Service (FaaS) being a widely used implementation. Nevertheless, due to the emergence of serverless platforms, it has become necessary to migrate microservices to serverless in order to create effective applications with streamlined infrastructure management, less operational burden, and cost savings. Furthermore, current research indicates that serverless solutions experience cold start latency. One recent study recommends the best programming language to use in serverless platforms which yields low cold start latency. Considering the those programming language, In this work, we try to compare the performance of serverless functions. Those functions that are implemented as serverless applications demonstrate enhanced response time and throughput. Ultimately we are going to discover that is there any impact of code size on cold start on serverless platforms. After empirical analysis, It is observed that there is no relation between code size and cold start.