Research On Deduplication-enabled Redundant Array Of Independent Solid-State Drives

In the last ten years,with the development of technology and improvement of productivity,solid-state drives(SSD),which is based on NAND flash,has gained rapid growth.Compared with hard disk drive(HDD),SSD has no mechanical parts.Thus,SSD can provide better read/write performance,resist higher shock,consume less power,and produce less noise than HDD.SSDs now become de facto primary storage plan for PC and laptops,and are widely used in enterprise data centers.To improve the capacity of storage system and ensure the data reliability,more and more enterprise users utilize disk array to store data.However,SSDs have different storage charactristics from traditional mechanical HDDs,SSDs are not feasible to update data in place and require regular garbage collection to restore data blocks.Simply applying the disk-array technology to SSDs will not take the performance advantages of SSDs,but may lead to write imbalance and performance fluctuation.On the other hand,the global data grows exponentially.In order to reduce the data storage pressure,the data deduplication technology emerged around 2000.The data deduplication technology can effectively reduce the data storage overhead,improve the write performance and save the network bandwidth.Currently,the data deduplication technology has been widely used in enterprise storage,cloud storage,and so on.This paper proposes a Deduplication-enabled Redudant Array of Independent SSDs architecture,short for DeRAIS,which utilizes the data deduplication technology to construct RAIS.By combining the data redundant information and data layout information which data deduplication brings,DeRAIS groups the data according to the different redundant rate,DeRAIS also processes the garbage collection operations on the strip level,to ensure stripe availability and solve performance fluctuation problem.The evaluation results show that DeRAIS solves the performance fluctuation problem caused by the unbalanced writes of RAIS and the garbage collection operations,and effectively makes full use of the advantages of RAIS,which have been verified in the tests conducted by different workloads.Moreover,the flexible design of DeRAIS enables it to be used in both software and hardware,thus achieving a good architecture.