基于非易失性存储器的低能耗数据存储方法研究

In recent years, the new nonvolatile memory (NVM) has become a hot issue in the distributed storage as its nonvolatile, high performance, low power consumption, and high scalability. However, the reads and writes of NVM are asymmetry, that is, the cost of writing operations is more than the cost of reading operations. It results in higher write energy consumption. Therefore, it is a very important issue that how to design an efficient data allocation and storage optimization algorithm based on NVM. The goal is to reduce energy consumption while ensuring storage performance and storage lifetime. This proposal focuses on the key problem of reducing energy consumption, and analyzes the problems existing in the previous works on low-power data allocation and storage optimization. In this proposal, firstly, the memory-access data flow graph (MDFG) model will be established and low energy consumption data distribution and storage optimization criteria will be designed, based on the information of NVM and application data. Secondly, to satisfy real-time performance requirements and balance performance improvement and energy consumption reduction, performance guarantee and low energy consumption data allocation and storage optimization algorithm will be designed. Finally, to achieve low energy consumption data allocation optimization under the guarantee of memory lifetime, a write-aware data allocation algorithm and data update strategy will be designed. The research of the subject will provide an important reference for the future storage management technology.

近年来，新型非易失性存储器（NVM）因其具有非易失、高性能、低功耗、高可扩展性等特点，成为近期分布式存储的热点研究问题。然而，NVM具有读写不对称性性，即写操作代价要比读操作代价高很多，从而导致较高的写能耗。因此，如何基于NVM设计高效的数据分布与存储优化算法来降低能耗同时保证存储性能、改善存储寿命是一个非常重要的问题。本课题着眼于“降低能耗”这一绿色存储的关键问题，首先分析现有低能耗数据分布与存储优化算法中存在的问题，基于NVM与应用数据个体信息，建立数据存储访问模型，并设计能耗有效的数据分布与存储优化准则；其次根据应用实时性等性能需求，为平衡性能改善与能耗降低，设计性能保证的低能耗数据分布与存储优化算法；最后，在保证存储寿命的前提下实现低能耗数据分布与存储，设计基于写感知的数据分布算法与数据更新策略。课题的研究为未来存储管理技术提供重要的借鉴。

近年来，新型非易失性存储器（NVM）因其具有非易失、高性能、低功耗、高可扩展性等特点，成为近期分布式存储的热点研究问题。然而，NVM具有读写不对称性性，即写操作代价要比读操作代价高很多，从而导致较高的写能耗。因此，如何基于NVM设计高效的数据分布与存储优化算法来降低能耗同时保证存储性能、改善存储寿命是一个非常重要的问题。本课题着眼于“降低能耗”这一绿色存储的关键问题，首先分析现有低能耗数据分布与存储优化算法中存在的问题，基于NVM与应用数据个体信息，建立数据存储访问模型，并设计能耗有效的数据分布与存储优化准则；其次根据应用实时性等性能需求，为平衡性能改善与能耗降低，设计性能保证的低能耗数据分布与存储优化算法，EADA和BDAEW和算法；最后，在保证存储寿命的前提下实现低能耗数据分布与存储，设计基于写感知的数据分布算法与数据更新策略。课题的研究为未来存储管理技术提供重要的借鉴。