作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2022, Vol. 48 ›› Issue (3): 54-59. doi: 10.19678/j.issn.1000-3428.0061274

• 人工智能与模式识别 • 上一篇    下一篇

基于数据更新间隔的NAND闪存垃圾回收算法

余进, 严华   

  1. 四川大学 电子信息学院, 成都 610065
  • 收稿日期:2021-03-25 修回日期:2021-05-27 发布日期:2021-06-03
  • 作者简介:余进(1998-),男,硕士研究生,主研方向为模式识别、智能系统;严华,教授、博士。
  • 基金资助:
    国家自然科学基金(11872069)。

Garbage Collection Algorithm for NAND Flash Memory Based on Data Update Interval

YU Jin, YAN Hua   

  1. College of Electronics and Information Engineering, Sichuan University, Chengdu 610065, China
  • Received:2021-03-25 Revised:2021-05-27 Published:2021-06-03

摘要: 闪存因具有速度快、体积小等优点而广泛应用于数据存储领域,为提高NAND闪存的垃圾回收效率、延长闪存使用寿命,提出一种基于数据更新间隔的垃圾回收算法UIGC。计算闪存中空闲页的分散度,将其作为垃圾回收触发条件。从垃圾回收效率和磨损均衡效果2个方面出发,综合考虑块中无效页年龄累计和以及块中有效页比例,使用动态回收块选择和静态回收块选择相结合的策略来选择目标回收块,根据回收块中有效页数据更新间隔判断有效页热度,同时提出数据更新稳定性的概念来划分有效页的数据更新状态,将具有不同热度和更新状态的有效页数据分别存储在不同的空闲块中,从而提高块中数据的同步更新概率。实验结果表明,UIGC算法相较于CAT、FaGC等现有垃圾回收算法具有更优的垃圾回收效率和磨损均衡效果,并能有效延长闪存使用寿命。

关键词: NAND闪存垃圾, 磨损均衡, 更新间隔, 数据热度, 更新稳定性, 数据分离, 文件系统

Abstract: Flash memory is widely used in the field of data storage, owing to its high speed and compact volume.In this study, we propose a garbage collection algorithm called UIGC based on data update interval to improve the garbage collection efficiency of NAND flash memory and prolong the service life of flash memory in general.The proposed approach calculates the dispersion of empty free pages in flash memory as the trigger condition for garbage collection. With the two design aims of garbage collection efficiency and wear balance performance, the proposed approach comprehensively considers the cumulative age sum of invalid pages and the proportion of effective pages in a block. Additionally, our approach combines dynamic and static collection block selection to select a target collection block and judge the effective page heat according to the data update interval of effective pages therein.Furthermore, the concept of data update stability is proposed, which is used simultaneously to divide the data update status of effective pages, and the effective page data with different heat and update statuses are stored in different idle blocks to improve the synchronous update probability of the data in the blocks.The experimental results show that the proposed UIGC algorithm exhibited better garbage collection efficiency and wear balance performance than existing garbage collection algorithms, such as CAT and FaGC, and that it can effectively prolong the service life of flash memory.

Key words: NAND flash memory garbage, wear leveling, update interval, data heat degree, update stability, data partition, file system

中图分类号: