传统网络数据分析的智能化升级IntelligentUpgradingofTraditionalNetworkTrafficAnalysisAIOps在企业数据中心的探索ExplorationofAIPOSinEnterpriseDataCenterAIOps三阶段•AlgorithmicITOperationsArtificialIntelligenceforITOperations•AutonomousITOperationsAIOps第一驱动力第二驱动力:IT架构MonolithicArchitectureTraditionalSOAMicroservices自动化分析-主动、高效性能管理解决方案业务运行态势感知,可视基线分析趋势预测RCA分析主动业务梳理业务关联分析全路径分析多参数组合阈值警报多段智能警报定制报表面向业务性能的时序数据库-CSPMDB应用协议解析,应用层交易日志130多种网络层到应用交易层性能指标(秒精度)CSPAE–科来流量分析处理引擎数据源-全网络数据包Flow数据分析结果警报DataSourceofAIOpsNetworkOSApplicationClientBusinessMachineLearningforAIOpsSupervisedLearningUnsupervisedLearningMachineLearningfromAliVisualizationtoOperationsOperationstoAutonomousOperationsHowtostartAIOpsNetworkOSApplicationClientBusinessSecurityNPMDSyslogAPMMobileSDKNPMDWhichisthebestBeginningAIOps技术成熟度实施周期资金成本团队能力安全WhyisNPMtheBestBeginning用户or前端FirewallADCADCWEBAPPADCCOREColaRASColaUPM技术成熟度资金成本实施周期生产安全团队能力GARTNERMagicQuadrant(2018)SituationalAwareness–BusinessSituationalAwareness–BranchsSituationalAwareness–ApplicationSituationalAwareness-NetworkSituationalAwareness-Event