Serverless Computing & Function-as-a-Service (FaaS) Optimization

Abstract

Function-as-a-Service (FaaS) in cloud computing is a critical optimization problem that needs to be tackled, including cold start latency, resource inefficiency, state management, and more. While FaaS provides obvious scalability and lower cost benefits, the lack of availability of resources and the problem of cold starts to prevent it from being used for high-performance applications. Pre-warming, snapshotting, and on-demand instantiation with lightweight runtimes, such as WebAssembly, are other ways to minimize cold start delays. It also benchmarks major FaaS platforms (AWS Lambda, Google Cloud Functions, Azure Functions, and OpenFaaS) and measures latency, throughput, and scalability metrics. The study also considers how to manage the resource, for instance, using auto-scaling, memory allocation, and request batching to enhance cost efficiency and performance. The study covers security challenges in multi-tenant environments and solutions for stateful applications in usually stateless serverless architectures with Faast.js, Knative, and OpenWhisk. Another area of research is edge computing and architectures for multiple clouds to improve the deployment of FaaS. Incorporating the lessons from this study gives it enough flexibility to adjust functions as applications in a real-world enterprise environment, especially in high-performance and data-sensitive applications. The study also provides security practices like function isolation and encryption to secure data in multi-tenancy environments for reliable, secure, and efficient serverless computing. The contribution to FaaS optimization and security for various use cases is achieved.