作者: Miriam Allalouf , Yuriy Arbitman , Michael Factor , Ronen I. Kat , Kalman Meth
关键词:
摘要: Power consumption is a major issue in today's datacenters. Storage typically comprises significant percentage of datacenter power. Thus, understanding, managing, and reducing storage power an essential aspect any efforts that address the total We developed scalable modeling method estimates workloads. The concept based on identifying workload contributors to consumed by disk arrays.To estimate given host workload, our translates primitive activities induced disks. In addition, we identified I/O queues have fundamental influence consumption. Our estimation results are highly accurate, with only 2% deviation for typical random workloads small transfer sizes (up 8K), up 8% large sizes. successfully integrated into power-aware capacity planning tool predict system requirements it online provide consumed.