Litoiu, MarinJiang, Zhen Ming (Jack)Ngo, Kim Long2022-03-032022-03-032021-112022-03-03http://hdl.handle.net/10315/39122Function as a Service (FaaS) is a new cloud technology where resource management is automatically handled by cloud providers. However, the automated resource management also reduces the transparency needed for software engineering tasks and additional FaaS' characteristics such as cloud function idle timeout, auto-scaling policies, response time to bursting workloads are unknown to software engineers. In this thesis, we propose a methodology to measure the cloud function instance idle timeout. Next, we characterize FaaS' scalability and elasticity using intensive workloads. Finally, we propose a strategy to improve the FaaS' performance under saturation scenario. The results show that cloud function instances are decommissioned if being left idle beyond certain period. Load and performance experiments reveal that different cloud platforms adopt distinct auto-scaling policies and when FaaS has reached the upper concurrency limit, a workload smoother can help to boost the system's success rates from 60 - 80% to 99 - 100%.Author owns copyright, except where explicitly noted. Please contact the author directly with licensing requests.Computer scienceElectronic Thesis or Dissertation2022-03-03ServerlessFunction as a serviceIdleTimeoutScalabilityElasticityWorkload smootherSoftwarePerformanceImprovement