Overview amp Roadmap Contact Srinivasaraddepalliintelcom 1 Traditional Cloud Native frameworks For Enterprise applications Linux K8s Linux K8s Public Cloud K8s Cluster VMs Private Cloud K8s cluster ID: 1043945
Download Presentation The PPT/PDF document "Edge (and cloud) and Multi-Cluster Orche..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
1. Edge (and cloud) and Multi-Cluster OrchestratorOverview & RoadmapContact: Srinivasa.r.addepalli@intel.com1
2. Traditional Cloud Native frameworksFor Enterprise applicationsLinuxK8sLinuxK8sPublic Cloud K8s ClusterVMsPrivate Cloud K8s clusterAppsAppsTraditional:Number of K8s clusters are smallK8s Cluster installation/upgrades are mostly done independently in each location. Deployment of applications on K8s clusters is also done independently.K8s clusters are used for normal applicationsNetwork and security functions are deployed outside of K8s clusters as physical appliances or virtual appliances, typically supplied by providers.Today K8s Clusters are not meant for Network functions. Need for comprehensive converged Edge platforms.FWSLBIDSProvider suppliedFWSLBIDSProvider suppliedInternet
3. DriversLatency/ PhysicsBandwidth/ EconomicsContext/ ProximityPrivacy/ Legal3Trend : Geo Distributed Computing trend with Edge-computingWANPublic/ Private CloudmS4mS4mS3mS2mS1mS1Cloud PlatformNetwork (LAN/ WAN)Edge 1mS1mS2mS1EdgePlatformWANPublic/ Private CloudmS4mS3mS4Cloud PlatformExternal SystemEdge nmS1mS2mS1EdgePlatformWAN
4. 4Geo-Distributed Computing - few use cases5G Use caseUniversal CPE Use caseLarge Number of sitesComputing (Apps across sites) – MECMultiple tenant applications along with operator CNFs.Workload types - VMs, VNFs, CNFs, CNAs and Functions (FaaS)Note: K8s is becoming choice of workload orchestrator in each clusterMulti Edge/Cloud computing scale is similar (or even higher) to Hyper-scalers’ scaleNow Telcos, MSPs and Enterprises need @scale Orchestration and Automation solutions Enterprise (Retail) use case
5. 5Edge Computing – Similar to Cloud-computing, but with some special needsUniform Developer Experience across Clouds and EdgesSupport for all kinds of workloads ( VM, Containers and Serverless functions)Easy migration of workloads among Edges and Clouds Resource Constraints (Power, Cost, Space)Converged Edge supporting IT, OT applications & Network functionsOptimized infrastructure softwareAccelerator usage (Hence, awareness without losing platform independence propertyEdge requires high security assurance (No physical security in far edges)Platform attestation ConfidentialityMulti-tenancy isolationEase-of-Use(@Scale requirements are higher than the clouds)Infrastructure Orchestration (K8s Cluster Life cycle management)Multi Cluster Distributed Application Orchestration & LCMMulti Cloud Uniform Networking (Overlay)5G Breakout5G UPFRAN Acceleration
6. E2E Edge StackLinuxK8sLinuxPublic Cloud K8s ClusterVMsPrivate Cloud K8s clusterLinuxEdge K8s ClusterPossibly in hundreds/thousandsDistributed Edge Orchestrator (Multi Cluster Orchestrator)Distributed Applications & Network functionsCI/CD workflowsMulti Cloud Networking HubsInfrastructure OrchestratorK8sK8sEdge Extensions (OpenNESS)Edge ExtensionsEdge Extensions
7. 7Service Orchestrator – Big PictureService OrchestratorOne Click deployment of complex applications & network services across multiple K8s clustersComprehensive Status monitoring of deployed complex applicationsOne Service Orchestrator for both CNFs and ApplicationsSelf Service Portal for multiple tenantsSingle pane of glass for Day0/Day1/Day2 configuration of CNFs and Apps.Comprehensive Analytics platform for Day2 operations
8. 8Needs/Requirements – PreparationService OrchestratorRegistration of Clusters Cluster labels(Example: Cell tower Edge, CO Edge etc..)Needed for identifying multiple clusters Cluster ProvisioningTenant ProvisioningLogical Cluster ProvisioningTenant registrationAbility to use tenant specific OAUTH2 servers for authenticating tenant adminsTenant level isolation via RBAC rulesLogical Cluster provisioning across multiple selected clustersLogical clusterLogical clusterLogical Cluster user and permission provisioningCluster specific configuration(Few: ISTIO CA provisioning; Virtual/Provider network preparation)
9. 9Needs/Requirements – Application deployment designService OrchestratorApp Onboarding(Complex Apps & Network Services)Multiple deployment profiles to ensure same APP can be instantiated multiple timesCluster ProvisioningTenant ProvisioningLogical Cluster ProvisioningPlacement constraints :Affinity and Anti-Affinity;Platform capabilities;Latency; CostTenant level isolation via RBAC rulesLogical Cluster provisioning across multiple selected clustersLogical clusterLogical clusterLogical Cluster user and permission provisioningPlacement policies to replicate and distribute workloads across clustersApp/Service onboardingDefine placement policiesDefine Placement constraints Logical cluster
10. 10Requirements – Workload Customization & Connectivity managementService OrchestratorNo changes to helm charts/K8s description of applicationsEach deployment may have its own customizationCluster ProvisioningTenant ProvisioningLogical Cluster ProvisioningConnectivity intent provisioningEnabling inter-micro service communication within or across clustersEnabling communication to external entitlesWith/Without Mutual TLSMulti Cluster DNS managementLogical clusterLogical clusterApp/Service onboardingDefine placement policiesDefine Placement constraints Logical clusterWorkload CustomizationAutomation of SM; SD-EWAN; firewall/NAT; SLB; DNSDynamic provisioning with LCM of ApplicationsExtensible framework to add new capability controllers
11. 11Requirements – OperationsService OrchestratorContinuous monitoring of complex Application(Across clusters, apps and micro-services)Comprehensive report on the application statusCluster ProvisioningTenant ProvisioningLogical Cluster ProvisioningAnalytics frameworkMetric collection across clusters & appsLong term central store (Time Series)Training frameworkClosed loop policy managementLogical clusterLogical clusterApp/Service onboardingDefine placement policiesDefine Placement constraints Logical clusterWorkload CustomizationAutomation of SM; SD-EWAN; firewall/NAT; SLB; DNSContinuous App monitoringAnalytics & Closed loop automationDay2 Config ControllersDay 2 ConfigurationConfiguration of apps/network-functions that are already deployed.Various types of configurations (CR based, RESTful based or Netconf/yang based)
12. 12EMCO – Edge Multi Cluster OrchestratorPlatformsEnterprise EdgesEdge CloudsTelco COEdgesPub/Pvt CloudsNetworkEdgesDistributed Application scheduler Hardware PlatformAware ControllerDistributed Cloud ManagerCluster Registration ControllerTraffic Connectivity ControllerResource Synchronizer & Status MonitoringCLI/GUIEMCOCluster Registration Controller registers clusters by cluster ownersDistributed Application Scheduler provides simplified, and extensible placement; tenant mgmt; LCM implementationHardware Platform Aware Controller enables scheduling with auto-discovery of platform features/ capabilitiesDistributed Cloud Manager presents a single logical cloud from multiple edgesTraffic Connectivity controller auto-configure service mesh (ISTIO) and security policy (NAT, firewall), DNS and SLB entities of edges.Day2 generic configuration configures Day2 configuration of any app/network function via templates & configs.Resource Synchronizer & Monitoring synchronizes resources across multiple edge/cloud platforms and then monitors the status of deployed resourcesEMCO is an implementation of Service OrchestratorAddressing majority of requirements; Extensible architecture allows new automation requirementsOpen source EMCO can be downloaded from here Day2Cfg Generic
13. 13EMCO IntegrationsPart of these commercial solutionsPart of AMCOP solution from Aarna: https://www.aarnanetworks.com/amcopTwo SIs are using this in their solutionsParts of it are in one 5G deploymentsFew Blueprints in LFE/Akraino use EMCO for Multi Cluster OrchestrationONAP uses EMCO to Onboard/design network services and deploy on K8s clustersONAP Slicing Orchestrator uses EMCO for Day 2 LCM
14. 14EMCO Roadmap 1)One-click deployment of “Distributed Applications & Network Functions” across multiple Kubernetes clusters2) Metrics collection and centralized store1) Scheduling with awareness of platform-capabilities2) Service Mesh Orchestration across Kubernetes clusters3) Upgrades/Updates4) Generic K8s resource customizationCentralized deployment of applications & NFs into multiple Kubernetes clusters1) Security Orchestration2) SD-EWAN Orchestration3) AI Analytics & Closed loop4) Resiliency with Uber Cadence5) Day2 configuration6) Analytics & Closed loop
15. 15Thank You!!
16. 16EMCO Vision Be a comprehensive geo-distributed Cloud native application orchestratorBe a Multi-Party and Multi-Cloud Orchestrator Be an orchestrator for Network services and Enterprise applicationsBe an orchestrator for convergence of Network services and Enterprise applicationsBe an orchestrator for Distributed Clouds with Edge-computing
17. 17EMCO Market Interest MarketUse caseInterestuCPE 2.0 & 3.0LCM of SD-WAN, Security CNFs with uCPE for Edge-computingTier1 MSP5GLCM of Disaggregated RAN (DU, CU-CP, CU-UP)LCM of 5GC with dUPFLCM of Private-5GFew Telco interestCNF Vendors interestSIs interestEnterpriseEnterprise Edges & LCM of Enterprise applicationsGood interestIndustry 4.0Robotic ArmCloud Native PCLCM of backend of applications on PC/Laptop