本文選題：云存儲　切入點：緩存替換　出處：《計算機工程與設(shè)計》2017年06期 　論文類型：期刊論文

【摘要】：針對云存儲中緩存的實時有效替換問題,提出一種基于多層感知器(MLP)神經(jīng)網(wǎng)絡(luò)和緩存對象混合價值計算的緩存替換方案。將采集的云存儲訪問數(shù)據(jù)進行預(yù)處理,利用LRU算法獲得一個初始k-LRU集合;采用k-LRU集的一部分訓(xùn)練MLP神經(jīng)網(wǎng)絡(luò),獲得最優(yōu)窗口大小參數(shù);根據(jù)該窗口大小,在考慮緩存對象的下載延遲、訪問頻率、剩余壽命和成本因素下,計算緩存對象的混合價值,將最低價值的對象進行替換。實驗結(jié)果表明,該方法能夠有效提高緩存命中率,降低訪問延遲和成本。
[Abstract]:According to the cache in the cloud storage real-time replacement problem is proposed based on multilayer perceptron (MLP) neural network and hybrid cache object value cache replacement scheme. The acquisition of cloud storage access data preprocessing, using LRU algorithm to obtain an initial k-LRU set; k-LRU set is used as part of the training of MLP neural network, get the optimal window size parameter; according to the window size, considering delay, Download cache object access frequency, residual life and cost factors, the calculation value of hybrid cache object, the object will replace the minimum value. The experimental results show that this method can effectively improve the cache hit rate and decrease the access delay and cost.

【作者單位】： 廣州工商學(xué)院計算機科學(xué)與工程系;廣東工業(yè)大學(xué)計算機學(xué)院;
【基金】：廣東省青年創(chuàng)新人才類基金項目(2015KQNCX196)
【分類號】：TP333
，

本文編號：1644295