Scalable Cold-Start Optimization in Serverless Computing: Leveraging Function Fusion With PanOpticon Simulator

Serverless computing has transformed cloud computing with its inherent scalability, pay-as-you-go pricing model, and low-latency execution. However, cold-start delays arising from the on-demand initialization of execution environments such as containers or virtual machines remain a persistent bottleneck, particularly for latency-sensitive applications. This paper addresses this challenge by proposing a novel function fusion approach to minimize cold-start latency in serverless environments. The proposed strategy leverages the PanOpticon simulator, a comprehensive platform for deploying and evaluating Function-as-a-Service (FaaS) applications. The proposed approach reduces redundant initialization overhead and improves resource utilization by dynamically merging functions within a workflow. Unlike conventional frameworks, our method integrates function fusion directly into the simulation and evaluation pipeline, enabling more scalable and performance-aware optimizations. Extensive simulation-based evaluations demonstrate that the fusion strategy significantly reduces cold-start delays while enhancing system throughput and responsiveness across diverse workloads. The results confirm that function fusion not only minimizes latency but also maximizes efficiency, making it a robust solution for real-time, scalable serverless applications. This work contributes a practical framework for cold-start mitigation and lays the groundwork for future research in optimizing FaaS-based cloud architectures.