In this paper, we present EcoLife, an innovative strategy that leverages multi-generation hardware to co-optimize carbon footprint and service time within the serverless environment. EcoLife extends Particle Swarm Optimization (PSO) to adapt to the variations in serverless computing for making keep-alive and execution decisions. Our experimental results show that EcoLife effectively reduces carbon emissions while maintaining high performance for function execution.
│SC24_eco-life/
├──one_day/
├──three_day/
├──carbon_intensity/
├──data/
├──motivations/
├──node/
├──optimizers/
├──results/
├──selected traces/