Serverless-at-the-Edge Cold Start: SLO-Safe, Cost-Optimal, Carbon-Aware Mitigation

Abstract

Serverless computing provides a flexible and scalable manner of execution for latency-sensitive applications; nevertheless, cold start delay still remains a critical difficulty, especially when dealing with SLOs (Service Level Objectives). for edge environments that are resource-constrained. Current cold start mitigation methods mainly optimize either latency or cost, usually ignoring their combined impact on both SLO compliance and greenhouse gas emissions. The paper suggests an SLO-safe, low-cost and environment-friendly cold start mitigation framework for serverless-at-the-edge platforms. The framework utilizes smart planning to choose appropriate strategies, such as pre-warming or snapshot reuse, in the light of constraints on end-to-end tail latency, the cost of execution, and current carbon intensity. Experimental tests on typical edge workloads demonstrate that cold start delays; SLO violation rates; costs of execution and carbon emissions are all materially reduced. This has confirmed the feasibility of sustainable serverless computing at the edge.