unlocking-scalability-and-simplifying-multi-cloud-management-with-karmada-and-pipecd-zhi-karmadarepipecdji-xi-hu-fa-lia-khanh-tran-cyberagent-inc-hongcai-ren-huawei.pdf

1、Unlocking Scalability and Simplifying Multi-Cloud Management with Karmada and PipeCDHongcai Ren,Huawei,Karmada CommunityKhanh Tran,CyberAgent,PipeCD CommunityAgendaWhy multi-clusterIntroduction of KarmadaIntroduction of PipeCDKarmada x PipeCD Integration demoAgendaWhy multi-clusterIntroduction of Ka

2、rmadaIntroduction of PipeCDKarmada x PipeCD Integration demoWhy multi-cluster Location Latency:Deploy apps as close to the customers as possible Data locality:Keep user data in-country Isolation Environment(e.g.dev,test,prod)Security isolation:sensitive data must be isolated Organization isolation:T

3、eams have different domains Reliability Blast radius:Apps incident in one cluster must not impact the whole system Scale:The apps is too big to fit in a single cluster And so on Multi-Cloud and Multi-Cluster Deployment Has Become a Common PracticeMore than 87%of enterprise respondents are using the

4、services of multiple cloud vendors at the same time.Cloud native technologies and the cloud market are maturing,and an era of programmable multi-cloud management services is coming.Numerous clustersComplex and repeated cluster configurationsCluster management varies from vendor to vendorFragmented A

5、PI access entriesScattered servicesApplication configuration differs across clustersCross-cloud service accessApplication synchronization between clustersRestrictions from clustersRestricted resource schedulingRestricted application availabilityRestricted auto scalingVendor lock-inService deployment

6、 dependencyLack of automatic failoverLack of neutral open-source,multi-cluster orchestration projectsChallenges to multi-cloud container cluster managementChallenges of Being Cloud-Native Multi-CloudAgendaWhy multi-clusterIntroduction of KarmadaIntroduction of PipeCDKarmada x PipeCD Integration demo