keynote-jd-clouds-large-scale-serverless-practice-app-management-and-elastic-scaling-on-karmada-zha-mao-jhuan-daepnanokarmadazha-zhi-fa-lia-re-xiaofei-wang-jdcloud-1.pdf

1、JDCloud cross-cluster large-scale application management practiceJDCloud2024XiaoFei WangJDCloudCloudNativeEJDCloud2024Table of ContentsDevelopment Road of JD Cloud Containers0103Federation Auto-scaling04Summary02Federation ClusterCONTENTSJDCloud2024Development Road of JD Cloud Containers01JD Contain

2、erization Development HistoryJDOS1.0，Around the applications2014-20162019-2021Earliest implementationLarge-scale hybrid cloud container cluster.Full containerization,Support for promotional guaranteeThe containerization is the most thorough.Application middleware,database,big data,mixed deployment,a

3、rtificial intelligence.JDOS2.0Archimedes schedulingconduct refined operations100%coverageCarrying the business of JD Retail,Logistics,Finance,Health,Industry and Insurance.CNCF Platinum MemberEfficiency improvement.Achieve the optimal solution of efficiency,cost and elasticity in multi-cluster manag

4、ement.Hybrid cloud productsMulti-cloud scheduling2022-NOWJDVesselHybrid cloud operating system productsFederation Cluster,Application cross-cluster managementLarge-scale elastic scaling across clustersReduce costs and improve disaster recovery capabilities Fault explosion radius is too large,affecti

5、ng the application of SLA.The application of cross-cluster auto-scaling is difficult,manual operation and maintenance.Everything around the application.The Unified resource scheduling Multi-group flexible scheduling strategy Cross-cluster elastic scaling One-click migrationClusterfailureManualoperat

6、ionDifficultyinmigrationDisasterrecoveryand multiactivitySingle-Cluster limitationMulti-cluster SchedulingMulti-cluster auto-scaling Multiple applications Cross-cluster networkMultiple and HAJDCloud2024Federation Cluster02Producthttps:/ development and enhancement based on KarmadaArchitectureFedKSer