Cache selecting method and system based on classification forecasting

Abstract

The invention discloses a cache selecting strategy based on classification forecasting. According to the strategy, objects are divided into two types including 'only one-time access' objects and 'more than one-time access' objects by using naive bayesian classification thinking. The 'only one-time access' objects are prevented from being placed in a cache, and the price for classification time and space based on naive bayesian is small. In a caching process, pictures which are not forecasted accurately are compensated through a historical information sheet, and the error probability is reduced. In addition, the forecasting weight of 'only one-time access' is dynamically adjusted according to real-time conditions of the cache, and therefore, the cache can be utilized effectively. Pre-judgment before a replacement algorithm is cached, and cache selecting strategy can be used with an optional cache replacing algorithm. While the complexity of the algorithm of the cache is not changed, the cache hit rate is increased, SSD writing is reduced, and the service life of SSD is prolonged. Meanwhile, machine learning thinking is guided into the cache, and a new consideration is provided for research of the cache algorithm.

Description

technical field [0001] The invention belongs to the field of computer data storage, and more particularly relates to a cache selection method and system based on classification prediction. Background technique [0002] Recently, flash-based solid-state drives (SSDs) have been widely deployed as caching devices to improve caching system performance. SSD based on flash memory is mainly NAND flash memory type, and its read and write functions are realized based on block erasing and writing. Compared with traditional mechanical hard drives, flash-based solid-state drives have the advantages of lower energy consumption, faster speed, stronger shock resistance, and higher stability, so this type of flash-based solid-state drives can Applied in various environments. [0003] Cache management algorithms for solid-state disks have been well studied in the past few decades, but they cannot be directly used to manage SSDs because of the following problems: First, traditional caching ...

AI Technical Summary

Problems solved by technology

[0004] In view of the above defects or improvement needs of the prior art, the present invention provides a cache selection method and system based on classification prediction. It will quickly wear out the SSD disk and increase the cost of using the SSD disk, as well as the technical problem that the cache utilization rate decreases due to the elimination of some hot data

Images