MCS-B: An Energy Efficient Storage System for Astronomical Observation Data Based on Logical Block Replacement Strategy
Chao Sun, Shanjiang Tang, Zichao Yuan, Ce Yu, Jian Kun Xiao, Jizhou Sun, Shuibing He · 2017
Power conservation for the storage system is crucial for cost saving and there are many approaches to address it, such as multi-level storage and data migration strategy. However, most of them are purely theoretical studies without real experiments. In realistic application scenarios, there can be a variety of factors that can impact the effectiveness of the energy efficient model. In this paper, a new energy efficient storage system for astronomical observation data called MCS-B is designed, which uses multi-level caching strategy and block replacement strategy. According to the manner of astronomical data access, MCS-B uses LB (logical blocks) to cluster correlated files and places them on the data disk via a temporal and spatial pattern. We further design a specialized prefetching and caching strategy to create larger disk idle time intervals so that the disks that are not in use within a certain time will be powered down. Experimental results show that MCS-B outperforms existing traditional energy efficient schemes in the energy consumption by 34.24% up to 42.39%. Moreover, the average response time for requests is reduced by 71.36% up to 77.14%.