ColdStart Reduction in Serverless Using Function Discovery and Distribution

Serverless, Function-as-a-Service (FaaS) is a cloud model that enables on-demand function execution with minimal management overhead for users, as cloud providers fully manage infrastructure. However, serverless computing suffers from a crucial drawback known as cold start latency, where initializing function containers from an idle state introduces substantial delays. Improving cold-start latency for serverless functions is essential for latency-sensitive applications, such as those used in 5G and 6G network infrastructures. This paper introduces a distributed middleware solution that reduces cold start latency by enabling function sharing and discovery across multiple nodes. We evaluated our solution using the OpenFaaS open-source platform, conducting experiments on a simulated network with varied latency conditions. Our middleware achieved a 80% reduction in cold start latency compared to a baseline OpenFaaS deployment without our middleware, demonstrating significant improvements in response times. Our approach minimizes container startup delays and improves resource efficiency by dynamically routing function requests to already initialized instances on neighboring nodes. These results indicate that our solution is well suited to enhance serverless applicability in next-generation mobile network environments, making serverless architectures viable for 5G Core networks and prospective 6G deployments.