基于队列计数的固态存储器热数据识别方法

Method about identifying hot data in solid state memory based on queue count

目前较常用的热数据识别方法主要关注于数据的访问频繁度,却没有有效地体现出访问次数随时间的变化关系,从而造成了较高的错误识别率。针对这种不足,设计了基于队列计数的热数据识别方法,通过使用先进先出规则淘汰过时的数据,并赋予队列中每个位置所存放元素不同的权值来表示不同时刻各数据的热状态;同时结合Flash读写等特性,引入了热区域概念,使其所需存储空间大大降低。通过理论分析以及实验证明,此方法在热数据识别和均衡效果方面达到了较好的性能,从而提高了垃圾回收的效率以及延长了存储器的使用寿命。

Now,commonly used recognition method of hot data is focusing on the degree of data access frequency,but it does not effectively reflect the changes between the number of visits with time and resulting in a higher error recognition rate.For this deficiency,this paper designed a hot data recognition algorithm based on queue count,by using the FIFO rule to eliminate the outdated data,and indicated the different hot state of different moments by giving a distinct weight to each element of the queue.Combined with the read and write features of the Flash,introduced the concept of hot area,and reduced the required storage space greatly.Theoretical analysis and experimental results show that this method achieves a better performance in hot data recognition and wear level,and then increases the efficiency of garbage collection and extends the life of the memory.