Method and system for ensuring data crash consistency in persistent memory

Abstract

The invention discloses a method and a system for guaranteeing consistency of data crash in a persistent memory, and belongs to the field of computer storage. The method comprise the following stepsof: dividing a system operation process into a plurality of continuous operation periods comprising an execution stage and a check point stage; in the current operation period Ri, for each processor core Cj, executing the workload according to an execution strategy of the processor core Cj in the operation period Ri in an execution stage, and marking the updated data as a work copy; in a check point phase, respectively writing the execution state of the processor core Cj and the working copy in the volatile cache into the persistent memory and the nonvolatile cache; marking the work copies ascheck point data, and determining an execution strategy of the processor core Cj in the next operation period according to the characteristics of the executed work load, thereby creating a new check point and ending the operation of the system in the current operation period. The expenditure for ensuring the data crash consistency in the persistent memory can be reduced.

Description

technical field [0001] The invention belongs to the field of computer storage, and more specifically relates to a method and system for ensuring data crash consistency in persistent memory. Background technique [0002] With the continuous development of computer systems, traditional dynamic random access memory (Dynamic Random Access Memory, DRAM) has been unable to meet the current system's demand for memory due to limitations such as refresh power consumption and process size. New non-volatile memory (Non-Volatile Memory, NVM), such as phase change memory (Phase Change Memory, PCM), spin transfer torque random access memory (Spin Transfer Torque Random Access Memory, STT-RAM), resistive memory (Resistive Random Access Memory, RRAM), etc., are considered as one of the substitutes for DRAM due to its low power consumption, non-volatile, byte addressable and other characteristics. There are currently many ways to integrate NVM into a computer system, one of which is to comp...

AI Technical Summary

Problems solved by technology

Therefore, the existing methods for ensuring data crash consistency in persistent memory will cause a large amount of data management overhead and checkpoint overhead when used in a pure persistent memory memory system, reducing the overall performance of the computer system

Images