Green or Fast? Learning to Balance Cold Starts and Idle Carbon in Serverless Computing

Researchers developed LACE-RL, a deep reinforcement learning framework that optimizes serverless computing by balancing cold-start latency and carbon emissions. The system dynamically adjusts keep-alive durations based on real-time carbon intensity and workload patterns, achieving 51.69% fewer cold starts and 77.08% lower idle carbon emissions compared to static policies.

• →LACE-RL uses deep reinforcement learning to dynamically optimize serverless pod retention decisions in real-time.
• →The framework reduces cold starts by 51.69% while cutting idle keep-alive carbon emissions by 77.08% compared to Huawei's static policy.
• →The system considers time-varying grid carbon intensity and workload patterns to make intelligent resource management decisions.
• →Performance testing used Huawei Public Cloud Trace data and approached Oracle-level performance benchmarks.
• →The research addresses the growing need for sustainable cloud computing practices in serverless architectures.