A Fault Detection and Diagnosis Approach for Multi-tier Application in Cloud Computing

Khiet Thanh Bui, Len Van Vo, Canh Minh Nguyen, Tran Vu Pham, and Hung Cong Tran

110 399.399-414

10.1109/JCN.2020.000023

Abstract :​​Ensuring the availability of cloud computing servicesalways concerns both service providers and end users.Therefore, the system always needs precautions for unexpectedcases. Accordingly, cloud computing services mustbe capable of identifying faults and behaving appropriatelywhen it is abnormal to ensure the smoothness as well asthe service quality. In this study, we propose a fault detectionmethod for multi-tier web application in cloud computingdeployment environment based on the Fuzzy Oneclasssupport vector machine and Exponentially WeightedMoving Average method. And then, the suspicious metricsare located by using feature selection method whichbased on Random Forest algorithm. To evaluate our approach,a multi-tier application is deployed by a transnationalweb e-Commerce benchmark by using TPC-W (TPCBenchmark™ W, simulates the activities of a business orientedtransaction web server in a controlled internet commerceenvironment) in private cloud and then it is injectedtypical faults. The effectiveness of the fault detection anddiagnosis are demonstrated in experiment results.​

Index terms :Cloud computing, fault detection, fuzzy One-class SVM, multi-tier web application