3732 - 统一 OCP 虚拟化集群的可观测性：一种可扩展的方法.pdf

1、Orlando,FLOctober 69IBM TechXchange 2025Explore.Build.Launch.Session 3732Jason EricksonIBM Technology Zone Platform Engineering ManagerUnifying Observability Across OCP Virtualization Clusters:A Scalable ApproachIBM TechXchange|2025 IBM Corporation2About MeAbout MeIBM Technology Zone Platform Engine

2、eringIBM Technology Zone Platform EngineeringPublic Cloud InfrastructureVirtualization PlatformsAutomationOur approach to monitoring our OpenShift Virtualization Clusters across 3 providers and with 11K+VMs in service.AgendaAgenda010203ChallengeSolutionBenefitsIBM TechXchange|2025 IBM CorporationCur

3、rent StateCurrent State28 OpenShift Clusters14 Hub Clusters14 Infrastructure Clusters14 IBM Storage Ceph Clusters11,000+Virtual Machines42 Prometheus InstancesClusters are not consolidatedDifferent networks,accounts,&providersIBM TechXchange|2025 IBM Corporation4OptionsOptionsPrometheus Remote Write

4、Prometheus Remote ReadThanosThats a lot of configuration.A lot of service IDs,tokens to rotate,etc.IBM TechXchange|2025 IBM Corporation5SolutionSolution1.Current Metrics2.Long Term Metrics3.Real Time Alerts4.mTLS Connectivity between servicesIBM TechXchange|2025 IBM Corporation6How It WorksHow It Wo

5、rksRed Hat Service InterconnectMutual TLS connection automatically managedExposes services between clustersThanos Querier:queries all cluster services Grafana DashboardIBM TechXchange|2025 IBM Corporation7ResultsResultsIBM TechXchange|2025 IBM Corporation8Unified DashboardSecurity without service Id

6、s and TokensLong Term metrics for trending and deeper analysisSupports multiple networks,locationsNow we have a centralized view into all metrics,secure,deployable with automation.IBM TechXchange|2025 IBM Corporation9Q&AJason EricksonIBM,Pl