程瑩 邵清
摘要:為了解決傳統(tǒng)服務(wù)器故障檢測方法大多針對已經(jīng)注入的故障類型進行檢測,無法獲取未知故障類型,并且檢測速度較慢的問題,提出一種基于自適應(yīng)監(jiān)測過程與決策樹算法的故障檢測方法ASFD。該算法利用自適應(yīng)監(jiān)測方法獲取服務(wù)器數(shù)據(jù),并引入信息熵與鄰居協(xié)作算法對故障檢測點進行檢測,然后將SVM與CART相結(jié)合進行故障類型判斷。實驗結(jié)果表明,該算法能夠有效實現(xiàn)故障類型判斷,提高了故障檢測速度。
關(guān)鍵詞:云計算;自適應(yīng);故障診斷;數(shù)據(jù)監(jiān)測;決策樹;SVM算法
DOIDOI:10.11907/rjdk.181088
中圖分類號:TP312
文獻標(biāo)識碼:A文章編號文章編號:16727800(2018)009007205
英文標(biāo)題Research on Adaptive Fault Diagnosis Algorithm for Server Fault in Cloud Environment
--副標(biāo)題
英文作者CHENG Ying, SHAO Qing
英文作者單位(School of OpticalElectrical and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 210093,China)
英文摘要Abstract:Most of the traditional server fault detection methods only detect the injected fault types,fail to get the unknown fault types,and the detection speed is slow.In order to solve this problem,a fault detection method based on adaptive monitoring process and decision tree algorithm is proposed ASFD in this paper.The algorithm uses adaptive monitoring method to get server data,and introduces information entropy and neighbor cooperation algorithm to detect the fault detection points.Then SVM and CART are combined to decide the fault type.The experimental results show that the algorithm proposed in this paper can effectively judge the fault type and enhance the fault diagnosis speed.
英文關(guān)鍵詞Key Words:cloud computing;selfadaptive;fault diagnosis;data monitoring;decision tree;SVM algorithm
0引言
隨著云計算發(fā)展不斷完善,用戶量增加導(dǎo)致服務(wù)器任務(wù)數(shù)量增加,且任務(wù)復(fù)雜性提高。云計算服務(wù)器負(fù)載增加及任務(wù)處理難度的提升使得服務(wù)器故障頻發(fā),易致系統(tǒng)崩潰[14]。
針對上述問題,眾多學(xué)者相繼提出了一系列算法用于云計算服務(wù)器故障診斷。文獻[5]將日志類型特征向量應(yīng)用于主故障與伴隨故障,提出基于伴隨狀態(tài)追蹤的持續(xù)故障定位框架CST,實現(xiàn)了注入故障類型檢測,該算法對已存在故障可以有效地判斷分析,但缺少對未知故障的分析。文獻[6]提出基于執(zhí)行軌跡監(jiān)測的故障診斷方式,采用代碼插樁監(jiān)測,然后利用主成分分析抽取關(guān)鍵方式診斷故障類型,該算法缺點主要是由于監(jiān)測方式與插樁方法數(shù)量成正比,因此性能消耗較大,不利于后續(xù)發(fā)展。……