Disk Energy saving strategy for Memory cloud architecture

Disk Energy saving strategy for Memory cloud architecture

Lu Liang Yu Yu Yingchang Sweet Wang Jinying Liu Mine

The advent of the Memory cloud (Ramcloud) improves the user experience of online data-intensive (Oldi) applications, but consumes more energy than traditional cloud data centers. To solve this problem, a disk energy-saving strategy for this architecture is proposed. Firstly, we introduce the fitness function and roulette method in genetic algorithm to choose more energy-saving disk for data persistent backup. Secondly, set reasonable server memory buffer to prolong the average continuous idle time of disk, make some server disk enter standby state when idle. The simulation results show that this strategy can save about 12.69% in the memory cloud system of 50 servers, and the buffer size setting has a double effect on energy saving effect and data availability.

Disk Energy saving strategy for Memory cloud architecture