Serverless computing in clouds (part A)

• Main Author: Li, Pingrui
• Other Authors: Tang Xueyan
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/165962
• Institution: Nanyang Technological University

Function as a Service (FaaS) is a cloud computing model that allows users to deploy self-contained functions without having to manage the underlying infrastructure. However, before a function can be executed, initialization work has to be done to prepare the runtime environment and the delay incurred is referred to as cold start overhead. To mitigate cold start overhead, containers can be kept alive and reused to execute subsequent invocations. The strategy of selecting containers to be kept alive or terminated is called keep-alive policy. Keep-alive policy resembles caching in many aspects and well-established caching algorithms can be used as a basis of designing keep-alive policies. This project proposes a few keep-alive policies based on existing caching strategies and evaluates the performance with functions of different characteristics. Public Microsoft Azure traces are used, and analysis is performed to have a better understanding of real-life workload in FaaS cloud systems.