Cloud Computing and Serverless Architectures Innovations and Applications
M. Vanisree, B. Srinivasarao, Seidakhmet Nurpatsha, Syed Riyaz Ahammed, Priyadarshini M C, E. Nagaraju · 2025
The paper provides integrated assessment for cloud computing and serverless architectures, demonstrating their performance, scalability and cost-effectiveness. We ran a standard set of experiments comparing traditional VM-based infrastructure to serverless platforms, including Chief as a Service (FaaS) models (such as AWS Lambda, Google Cloud Functions, and Azure Functions). Our end-to-end-latency measurements show a good RTT time of 60 ms at server less in average, while a VM-based setup spends 120 ms in average. But cold start latency of serverless functions was 300 ms averagely, still unacceptable for those latency-sensitive applications. Throughput and scalability benchmarks showed that serverless patterns per-form better and scale near linearly as load on the system increases, whereas VM-based ones suffer serious performance degradation at a certain load level. The cost analysis was used to show the huge financial benefits of serverless architecture cost being almost$\mathbf{5 0 \%}$less than the VM aggregates literal money. The model of serverless functions being billed on a pay-per-use basis and their capacity to prevent resource waste were to be very efficient for fluctuating workload applications. Application case studies on EC2 ecommerce platform, and data processing pipelines have further demonstrated the advantages of serverless architectures in providing reduced operating costs, and adaptiveness to workload variations. There are some issues, such as cold start latency, but otherwise serverless architectures offer some great benefits for modern applications. This report finds that serverless architectures are much better in terms of performance, scalability and costeffectiveness with respect to traditional cloud computing based on VM. Further research can investigate hybrid models and cold start latency remediation of serverless for all applications.