Energy-Aware Interval Caching for Mirrored Video Servers

Energy consumption is an important issue for data centers, and disks use a significant proportion of the total energy. Mirrored video servers handle data requests adaptively from their primary or backup copies allowing some disks to go into standby mode, and disk bandwidth utilization has a great effect on the total energy consumption. We analyze the relationship between the amount of cached data and the power consumption of each disk, and propose a new energy-aware interval caching (EAIC) scheme which considers the power state of the disks, so as to reduce their total energy consumption. Simulation results show that EAIC can save as much as 22% more energy than a conventional interval caching scheme.