Energy-aware workload scheduling in snowflake for sustainable big data computing

With rising concerns over cloud energy consumption, this research proposes a novel energy-aware workload scheduler for Snowflake's virtual warehouses. The study integrates energy-efficiency metrics into Snowflake’s resource provisioning mechanisms, aiming to minimize the environmental footprint of Big Data queries. Using a dataset of 10 million historical job runs, the scheduler predicts compute demands using LSTM-based time series models and defers non-urgent workloads to periods of lower grid carbon intensity. Simulation results show a 35% reduction in carbon footprint with only a 5% increase in average job latency. The scheduler also supports Snowflake’s multi-cluster auto-scaling and adapts dynamically to CPU utilization and I/O bursts. Case studies in retail analytics and IoT monitoring validate the practicality of the approach in real-world scenarios. The authors also propose energy dashboards embedded in Snowflake’s UI to promote transparency and green decision-making. This paper contributes to the emerging field of sustainable data warehousing by demonstrating how environmental goals can align with business intelligence, setting a precedent for ESG-compliant cloud analytics.