Abstract: Improving the efficiency of flash memory garbage collection and prolonging the life time of flash-memory can be achieved by finding out hot data. A framework of hierarchical hot data identification is proposed, which can find out hot data from large data set by recording the hot data information at several different levels. Experimental result shows this method can identify and predict hot data from large data set with restricted resources.

Key words: flash memory, hot data identification, large-scale

摘要： 热数据识别对于提高闪存垃圾回收的效率、延长闪存寿命都有重要的作用。针对大数据量下热数据识别难的情况，该文提出一种层次型热数据识别框架，采用在多个层次上记录热数据的方法。实验证明，该框架能够在消耗较少资源的情况下，识别和预测出大数据量中的热数据。

关键词: 闪存, 热数据识别, 大容量

CLC Number: 

• TP311