Serverless Architectures and Function-as-a-Service (FaaS): Scalability, Cost Efficiency, and Security Challenges

Abstract

Serverless computing and Function-as-a-Service (FaaS) have become the new “new hotness” in cloud computing. Since FaaS completely abstracts away the infrastructure, developers are left with having to write only their app logic — a very cost-effective and scalable solution for modern workloads. This paper looks at the rise of serverless architectures through these key platforms. We get answers on the benefits of the scalability and elasticity models and cost implications compared with traditional server-based infrastructures. The case study demonstrates how the pay-per-execution model can prevent resource waste and some problems developers might face in implementing this way of working, such as cold-start latency and informal ecosystems. We further present three case studies on serverless machine learning workloads, which show the practical benefits of serverless computing.