CloudCons: A Comprehensive End-to-End Benchmark for Cloud Resource Consolidation

CloudCons is a newly proposed benchmark aimed at evaluating forecasting models for cloud resource consolidation, addressing the limitations of existing benchmarks that focus only on prediction error metrics. The benchmark utilizes high-quality datasets from major cloud providers, including Huawei Cloud, Microsoft Azure, and Google Borg, and assesses various models, revealing that while foundation models excel in zero-shot forecasting accuracy, this does not necessarily improve decision utility. This research offers critical insights and guidelines for practitioners to optimize resource efficiency while maintaining service reliability in cloud environments.