ADL-RAID: Energy-Saving Data Layout for Dynamic Loads

Abstract

With the rapid development of information technology and the explosive growth of data, the scale of storage equipment is growing rapidly. The high energy consumption storage devices have become a serious problem for data centers. For data storage, the model of sequential data access a storage system is commonly used. Aiming at the characteristics of a sequential data storage system, this paper, we propose a structure, named Adapted to Dynamic Load based on Redundant Array Independent Disks (ADL-RAID) which is an effective energy-saving data layout for dynamic loads based on the existing Semi-Redundant Array Independent Disks (S-RAID). ADL-RAID inherits the local parallel energy-saving strategy, uses an address mapping mechanism, and allocates storage space to satisfy the performance requirements for the user requested application. By sensing different loads, ADL-RAID allocates storage space for its dynamic loads with the appropriate parallelism. One or several data disks are used when the load is minimized, and all the data disks are used in parallel when the load is maximized. Experimental results show that, for 100% continuous write request, ADL-RAID saves 33.6% energy consumption than S-RAID5 and improves write performance than S-RAID5 by 34.3%. Thus, ADL-RAID has higher availability and is ideal for sequential data storage applications