Serverless Computing for Small-Scale Applications: A Critical Review of Cost, Performance, and Deployment Trade-offs

Serverless computing lets developers run application code without having to manage the servers directly. Its event-driven design, automatic scaling, and usage-based billing can make it a useful option for small applications whose traffic rises and falls. However, serverless is not always the cheapest or fastest choice. Cold starts, execution limits, data-transfer fees, dependence on provider-specific services, monitoring requirements, and consistently high workloads can all affect the final outcome. This paper reviews research on serverless computing for small-scale applications, focusing on cost, performance, scalability, and operational effort. It uses a focused narrative review of selected surveys and studies on cold starts and cost, mainly published between 2021 and 2026, with earlier foundational work. Based on this literature, the paper develops a qualitative framework for comparing Function-as-a-Service (FaaS), virtual machines (VMs), and container-based services across different workload patterns. No new experiments are reported, and no single platform is presented as the best choice for every application. Instead, the review argues that developers should consider how predictable the workload is, how often functions run, how much delay users can tolerate, whether the application needs persistent state, and what the supporting services cost. The framework is an initial guide for the students and developers; before production use, it should be checked against the selected provider, current prices, and the application's actual requirements.