网络和云边缘参考系统架构产品组合用户手册

UserManual|NetworkandEdgeReferenceSystemArchitecturesPortfolio

2

TableofContents

1Introduction 1

1.1Purpose 1

1.2AudienceandScope 1

1.3PortfolioArchitecture 5

1.3.1KeyTerms 6

1.3.2IntelInvestmentsofCapabilities 7

1.4DeploymentProcess 8

2ReferenceSystemArchitecturesOverview 8

2.1DeploymentOptions 8

2.2KeyElements 11

2.3ConfigurationProfiles 12

2.3.1ConfigurationProfilesOverview 12

2.3.2On-PremisesEdgeConfigurationProfile(CustomerPremises) 12

2.3.3AccessEdgeConfigurationProfile(FarEdge) 13

2.3.4RemoteCentralOffice-ForwardingConfigurationProfile(NearEdge) 13

2.3.5RegionalDataCenterConfigurationProfile(RegionalPOP) 13

2.3.6Non-Location-SpecificConfigurationProfiles 13

2.4AnsiblePlaybooks 13

3HardwareComponents 14

3.14thGenIntelXeonScalableProcessorCapabilities 16

4SoftwareComponents 18

4.1KubernetesFeatures 18

4.2PlatformSystemFeatures 18

4.3ContainerRuntimes 19

4.4KubernetesPlugins 20

4.4.1MultusCNI 20

4.4.2SR-IOVNetworkDevicePlugin 20

4.4.3SR-IOVCNI 20

4.4.4UserspaceCNI 20

4.4.5BondCNI 20

4.4.6Intel®QuickAssistDevicePlugin 21

4.4.7Intel®SoftwareGuardExtensions(Intel®SGX)DevicePlugin 21

4.4.8NodeFeatureDiscovery 21

4.4.9TopologyManager 21

4.4.10KubernetesNativeCPUManager 22

4.4.11CPUControlPlaneManagement 22

4.4.12PlatformAwareScheduling 22

4.5IstioServiceMesh 23

4.5.1BypassTCP/IPStackinIstioServiceMeshUsingeBPF 23

4.5.2TLSSplicing 23

4.5.3IstioAccelerationwithIntelAVX-512Crypto 23

4.5.4IstioAccelerationandCompressionwithQAT2.0 23

4.5.5IstiomTLSPrivateKeyProtectionwithSGX 23

4.5.6IstioMulti-tenancy 24

4.6LinkerdServiceMesh 24

4.7KubernetesOperators 24

4.7.1SR-IOVNetworkOperator 24

4.7.2IntelDevicePluginsOperator 24

4.7.3MinIOOperator 24

4.7.4Intel®EthernetOperator 25

4.7.5ForwardErrorCorrection(FEC)Operator 26

4.8DynamicDevicePersonalization(DDP) 26

4.8.1DDPonIntelEthernet700SeriesNetworkAdapters 26

4.8.2DDPonIntel®Ethernet800SeriesNetworkAdapters 27

UserManual|NetworkandEdgeReferenceSystemArchitecturesPortfolio

3

4.9KubernetesPowerManager 28

4.10Intel®SpeedSelectTechnology 28

4.10.1IntelSpeedSelectTechnology–BaseFrequency 28

4.10.2Intel®SpeedSelectTechnology–CorePower 29

4.10.3Intel®SpeedSelectTechnology–TurboFrequency(Intel®SST-TF) 29

4.10.4Intel®SpeedSelectTechnology–PerformanceProfile(Intel®SST-PP) 29

4.11Security 29

4.11.1NativeKubernetesClusterSecurityFeatures 29

4.11.2Intel®SecurityLibrariesforDataCenter(Intel®SecL–DC) 30

4.11.3Intel®SoftwareGuardExtensions 30

4.11.4KeyManagementReferenceApplicationwithIntel®SGX 30

4.11.5OpenSSLandQATEngine 31

4.11.6TrustedCertificateService 31

4.11.7TrustedAttestationController 31

4.12Observability 31

4.12.1TelemetryInsightReports 32

4.13Storage 33

4.13.1MinIO 33

4.13.2LocalPersistentVolumeStaticProvisioner 33

4.13.3Rook/Ceph 33

4.14CloudNativeDataPlane 34

4.15ApplicationDeviceQueues(ADQ)Plugin 34

4.16TADK 34

4.17AccessEdge-vRAN 35

4.17.1GenericvRAN 35

4.17.2FlexRAN™SoftwareDeployment 35

5ReferenceArchitectureInstallation 35

5.1Introduction 35

5.2InstallationOverview 36

5.2.1OperatingSystemSelection 36

5.2.2NetworkInterfaceRequirements 36

5.2.3SoftwarePrerequisitesforAnsibleHost,ControlNodes,andWorkerNodes 36

5.3AnsiblePlaybookReview 36

5.3.1AnsiblePlaybookPhases 39

5.4DeploymentUsingAnsiblePlaybooks 39

5.4.1PrepareTargetServers 39

5.4.2GetAnsiblePlaybookandPrepareConfigurationTemplates 40

5.4.3UpdateAnsibleInventoryFile 40

5.4.4UpdateAnsibleHostandGroupVariables 40

5.4.5RunAnsibleClusterDeploymentPlaybook 40

5.4.6AnsibleClusterRemovalPlaybook 40

AppendixATerminology 41

AppendixBReferenceDocumentation 45

Figures

Figure1.LandscapeforNetworkandEdgeReferenceSystemArchitecturesPortfolio 5

Figure2.NetworkandEdgeReferenceSystemArchitecturesPortfolioHierarchy 5

Figure3.ConfigurationProfileImplementationExamples 6

Figure4.ContainerBareMetalReferenceSystemArchitecture(BMRA) 9

Figure5.VMRA–VirtualClusteronaSingleNode 9

Figure6.VMRA–VirtualClusteronMultipleNodes 10

Figure7.CloudRAforAmazonWebService(AWS)EKS 10

Figure8CloudRAforMicrosoftAzureKubernetesService 11

Figure9.ReferenceArchitectureKeyElements–BMRAExample 12

Figure10.BMRAAnsiblePlaybookOverview 14

Figure11.VMRAAnsiblePlaybookOverview 14

UserManual|NetworkandEdgeReferenceSystemArchitecturesPortfolio

4

Figure12.GenericKubernetesCRIStack 19

Figure13.DeprecatedKubernetesRuntimeStackUsingDocker 20

Figure14.StorageNodeDeploymentExample 25

Figure15.IntelEthernetOperatorArchitectureOverview 26

Figure16.CPUCoreFrequencyDeploymentMethods 28

Figure17.KeyManagementReferenceApplicationInfrastructurewithIntel®SGX 31

Figure18.TelemetryInsightReportsHigh-LevelArchitecture 33

Figure19.CloudNativeDataPlaneArchitecture 34

Figure20.TADKEnhancedImplementationofModSecurityWAF 35

Figure21.BMRAAnsiblePlaybooks–Build-Your-OwnConfigurationProfileExample 38

Figure22.VMRAAnsiblePlaybook–Build-Your-OwnConfigurationProfileExample 38

Tables

Table1.TermsUsed 6

Table2.HardwareandSoftwareConfigurationTaxonomy 7

Table3.IntelCapabilitiesInvestmentsandBenefits 7

Table4.HardwareOptionsforControlNode–4thGenIntelXeonScalableProcessor 15

Table5.HardwareComponentsforWorkerNodePlus–4thGenIntelXeonScalableProcessor 15

Table6.HardwareComponentsforWorkerNodePlus(AccessEdge-vRAN)–4thGenIntelXeonScalableProcessor 15

Table7.HardwareComponentsforWorkerNode(AccessEdge-vRAN)–4thGenIntelXeonScalableProcessorwith

IntegratedIntelvRANBoost 16

Table8.HardwareComponentsforStorageNode–3rdGenIntelXeonScalableProcessor 16

Table9.SiliconCapabilitiesof4thGenIntelXeonScalableProcessor 16

Table10.DDPDefaultPackageProfiles 27

Table11.Terminology 41

Table12.ReferenceDocuments 45

DocumentRevisionHistory

ThreepreviouseditionsoftheBMRAdocumentwerereleased,startinginApril2019.

1.Covered2ndGenIntel®Xeon®Scalableprocessors

2.Covered2ndand3rdGenIntel®Xeon®Scalableprocessors

3.Covered2ndand3rdGenIntel®Xeon®ScalableprocessorsandIntel®Xeon®Dprocessor

Revision

Date

Description

001

February2022

Initialrelease.

002

March2022

UpdatedafewURLs.

003

May2022

Coversthe4thGenIntel®Xeon®Scalableprocessor.

004

June2022

ThechangesincludeupdatestothediscussionoftheBMRAforStorageDeploymentModelandFigure2.

005

June2022

UpdatedafewURLs.

006

July2022

AddedNDAsupportforFlexRAN™software,updatedIstioandservicemeshfeatures.

007

July2022

ThechangesincludeupdatestothediscussionoftheAccessEdgeConfigurationProfileandtoFigures1and3.

008

October2022

AddedsupportforCloudReferenceSystemArchitecture(CloudRA).

009

December2022

IncludesimprovementsandupdatesonRAinalignmentwiththelaunchofthe4thGenIntel®Xeon®Scalableprocessor.

010

March2023

UpdatedforBMRARelease23.02;includesimprovementstorunFlexRAN™softwareinacontainerandadditionofMediaAnalyticsLibraries.

UserManual|NetworkandEdgeReferenceSystemArchitecturesPortfolio

5

1.3PortfolioArchitecture

Throughoutaninfrastructure—whetheraprivateenterpriseorpubliccloud—variousoptimizedsystemsareneededbasedontheworkloadsbeingrunandthesitesserviced

(Figure1)

.TheNetworkandEdgeReferenceSystemArchitecturesPortfolioprescribesdeploymentsolutionsoptimizedtosupportservicesbasedonthelocationintheinfrastructureandthetypical

applications/workloadsservedatthatlocation.Thesesolutionscanbedeployedquicklyusingvalidated,best-known

configurations(BKCs)andautomateddeploymenttoolsprovidedintheReferenceSystemArchitectures.Theconfigurationsandtoolsweredesignedforthenetworklocationandachosendeploymentmodel(containerizedorvirtual)todelivera

predictivesolutionforperformance-,security-,usability-,andcost-optimizedoperations.

Figure1.LandscapeforNetworkandEdgeReferenceSystemArchitecturesPortfolio

ThearchitecturalhierarchyfortheNetworkandEdgeReferenceSystemArchitecturesPortfolioisshownin

Figure2.

The

VMRAandBMRAimplementConfigurationProfilesthatprovidearecipeforimplementinghardware,software,andoptimizedconfigurationperthenetworklocation.

Figure3

illustratesanexampleofConfigurationProfilesdesignedfornetworklocations.

Figure2.NetworkandEdgeReferenceSystemArchitecturesPortfolioHierarchy

UserManual|NetworkandEdgeReferenceSystemArchitecturesPortfolio

6

Figure3.ConfigurationProfileImplementationExamples

1.3.1KeyTerms

Table1

liststhekeytermsusedthroughouttheportfolio.ThesetermsarespecifictoNetworkandEdgeReferenceSystemArchitecturesPortfoliodeployments.

Table1.TermsUsed

Description

Term

ExperienceKits

Guidelinesdeliveredintheformofmanuals,userguides,applicationnotes,solutionbriefs,and

trainingvideosforbest-practiceimplementationofcloudnativeandKubernetestechnologiestoeasedevelopmentsanddeployments

NetworkandEdgeReferenceSystemArchitecturesPortfolio

Atemplatedsystem-levelblueprintforarangeoflocationsinenterpriseandcloudinfrastructurewith

automateddeploymenttools.TheportfoliointegratesthelatestIntelplatformsandcloud-nativetechnologiesformultipledeploymentmodelstosimplifyandacceleratedeploymentsofkey

workloadsacrossaserviceinfrastructure.

DeploymentModel

ProvidesflexibilitytodeploysolutionsaccordingtoITneeds.Theportfoliooffersthreedeploymentmodels:

.ContainerBareMetalReferenceSystemArchitecture(BMRA)–AdeploymentmodelofaKubernetesclusterwithcontainersonabaremetalplatform.

.VirtualMachineReferenceSystemArchitecture(VMRA)–Adeploymentmodelofavirtualclusteronaphysicalnode.ThevirtualclustercanbeaKubernetescontainers-basedcluster.

.CloudReferenceSystemArchitecture(CloudRA)–AdeploymentmodelthatusesaCSP’sIntel-basedinstancesforrunningcloud-nativeapplicationsinthecloud.Theworkerinstancesare

providedbasedontheConfigurationProfilethatworkloaddemands.

ConfigurationProfiles

Aprescribedsetofcomponents—hardware,softwaremodules,hardware/softwareconfigurationspecifications—designedforadeploymentforspecificworkloadsatanetworklocation(suchas

AccessEdge).ConfigurationProfilesdefinethecomponentsforoptimizedperformance,usability,andcostpernetworklocationandworkloadneeds.

1

Inaddition,genericConfigurationProfilesareavailablefordevelopers’flexibledeployments.

ReferenceArchitectureFlavor

AReferenceArchitecturedeploymentusingaConfigurationProfile.

AnsiblePlaybook

Asetofvalidatedscriptsthatprepare,configure,anddeployaReferenceArchitectureFlavorbyimplementingaConfigurationProfile.

ConfigurationProfileAnsibleScripts

Automatequick,repeatable,andpredictivedeploymentsusingAnsibleplaybooks.Various

ConfigurationProfilesandAnsiblescriptsallowautomatedinstallationsthatareapplication-ready,dependingontheworkloadandnetworklocation.

1Workloadsandconfigurations.

Resultsmayvary.

UserManual|NetworkandEdgeReferenceSystemArchitecturesPortfolio

7

Term

Description

KubernetesCluster

Adeploymentthatinstallsatleastoneworkernoderunningcontainerizedapplications.Podsarethecomponentsoftheapplicationworkloadthatarehostedonworkernodes.Controlnodesmanagethepodsandworkernodes

IntelPlatforms

PrescribesIntelplatformsforoptimizedoperations.Theplatformsarebasedon3rdand4thGenIntel®Xeon®Scalableprocessors,Intel®Xeon®Dprocessor,and4thGenIntel®Xeon®ScalableprocessorwithvRANBoost.TheseplatformsincludetheTaylorsFallsReferenceDesign.The

platformsintegrateIntel®EthernetController700Seriesand800Series,Intel®QuickAssist

Technology(Intel®QAT),Intel®ServerGPU(GraphicProcessorUnit),Intel®DataCenterGPUFlexSeries,Intel®Optane™technology,andmore.

Note:ThisreleaseofVMRAdoesnotsupporttheIntelXeonDprocessor.

Inadditiontokeyterms,portfoliodeploymentproceduresfollowahardwareandsoftwareconfigurationtaxonomy.

Table2

describesthetaxonomyusedthroughoutthisdocument.

Table2.HardwareandSoftwareConfigurationTaxonomy

Term

Description

HardwareTaxonomy

ENABLED

SettingmustbeenabledintheBIOS(configuredasEnabled,Yes,True,orsimilarvalue)

DISABLED

SettingmustbedisabledintheBIOS(configuredasDisabled,No,False,oranyothervaluewiththismeaning)

OPTIONAL

Settingcanbeeitherdisabledorenabled,dependingonworkload.SettingdoesnotaffecttheConfigurationProfileorplatformdeployment

SoftwareTaxonomy

TRUE

Featureisincludedandenabledbydefault

FALSE

Featureisincludedbutdisabledbydefault-canbeenabledandconfiguredbyuser

N/A

Featureisnotincludedandcannotbeenabledorconfigured

1.3.2IntelInvestmentsofCapabilities

IntelinvestmentsinnetworkingsolutionsaredesignedtohelpITcentersacceleratedeployments,improveoperationalefficiencies,andlowercosts.

Table3

highlightsIntelinvestmentsintheportfolioandtheirbenefits.

Table3.IntelCapabilitiesInvestmentsandBenefits

Benefit

Capability

Performance

Intel®platforminnovationandaccelerators,combinedwithpacketprocessinginnovationforcloud-nativeenvironments,deliversuperiorandpredictiveapplicationandnetworkperformance.

2

OrchestrationandAutomation

ImplementingKubernetescontainersorchestration,includingKubernetesOperators,simplifiesandmanagesdeploymentsandremovesbarriersinKubernetestosupportnetworkingfunctionality.

Observability

Collectingplatformmetricsbyusing,asanexample,thecollectddaemonandTelegrafserveragent,publishingthedata,andgeneratingreports,enableshighvisibilityofplatformstatusandhealth.

PowerManagement

LeveragingIntelplatforminnovation,suchasIntel®SpeedSelectTechnology(Intel®SST),supportsoptimizedplatformpowerutilization.

Security

Intelsecuritytechnologieshelpensureplatformandtransportsecurity.Thesetechnologiesincludingthefollowing:

.Intel®SecurityLibrariesforDataCenter(Intel®SecL-DC)

.Intel®QuickAssistTechnologyEngineforOpenSSL(Intel®QATEngineforOpenSSL)

.Intel®SoftwareGuardExtensions(Intel®SGX)

.KeyManagementReferenceApplication(KMRA)implementation

Storage

Thefollowingstorageconfigurationsaresupported:

.Objectstorage-adisaggregated,high-performance,scalablestorageplatformusingMinIOObjectStoragesupportsdata-intensiveapplications,suchasmediastreaming,bigdataanalytics,AI,andmachinelearning.

.Localblockstorage-forcontainerizedworkloadsthatimplementLocalPersistentVolumeStaticProvisioner(LPVSP)forreliablecontainerattachedstoragebyturninganystorageavailableonKubernetesworkernodesintolocalordistributedpersistentvolumes(PVs).

.Rook/Ceph–Cephisascalabledistributedstoragesystemthatprovidesaunifiedstorageservicewithobject,block,andfileinterfacesfromasinglecluster.Rookturnsdistributedstoragesystems

2Workloadsandconfigurations.

Resultsmayvary.

8

Capability

Benefit

intoself-managing,self-scaling,self-healingstorageservicesandautomatesthetasksofastorageadministrator:deployment,bootstrapping,configuration,provisioning,scaling,upgrading,

migration,disasterrecovery,monitoring,andresourcemanagement.

ServiceMesh

ImplementingaServiceMesharchitectureusingIstioallowsapplicationservicesthatcanbeadded,connected,monitored,moresecure,andload-balancedwithfewornocodechanges.ServiceMeshisintegratedwithTrustedCertificateServiceforKubernetes*platform,providingsecurekey

management

1.4DeploymentProcess

Thisusermanualprovidesthefoundationalinformationfordeployments.Aftertheconceptshereareunderstood,refertooneofthefollowing,dependingontheintendeddeploymentscenario.

.Kubernetescontainersdeployments:

NetworkandEdgeContainerBareMetalReferenceSystemArchitectureUserGuide

.Virtualizedsystemdeployments:

NetworkandEdgeVirtualMachineReferenceSystemArchitectureUserGuide

.Clouddeployments:

NetworkandEdgeCloudReferenceSystemArchitectureUserGuide

Additionally,IntelprovidesExperienceKitsthatdetailthetechnologiesenabledintheReferenceSystemArchitectures

Portfolio,includingbenchmarkinformation.ExperienceKitsareavailableonIntel®NetworkBuildersat

Network&Edge

PlatformExperienceKits.

ForNDAmaterial,contactyourlocalIntelrepresentative.

2ReferenceSystemArchitecturesOverview

2.1DeploymentOptions

TheNetworkandEdgeReferenceSystemArchitecturesPortfoliooffersthreedeploymentoptions:BMRA,VMRA,andCloudRA.Theseoptionsaredesignedfortheconvergenceofkeyapplicationsandservices,controlplane,andhigh-performance

packetprocessingfunctionsinaninfrastructure.Toolssuchastestpmdandpktgenpodsandsamplecloudnativenetworkfunction(CNF)workloads(vCMTSandvBNG)areusedforvalidationoftheReferenceSystemArchitecturesdelivered.

TheBMRAisascalable,openKubernetesclustercomposedofphysicalworkerandcontrolnodesnetworkedthroughswitchesandconnectedasdepictedin

Figure4.

TheVMRAisascalablevirtualclustersupportingbothavirtualKubernetesclusterandaVMRAclusterwithascalablenumberofVMs.TheVMsareconnectedasavirtualclusterofworkerandcontrolVMswithinthesinglenodeasdepictedin

Figure5

andonmultiplenodesasdepictedin

Figure6.

TheCloudRAisascalableclusteronapubliccloud.Itsarchitectureisshownin

Figure7.

TheCloudRAprovidesthemeanstodevelopanddeploycloud-nativeapplicationsinaCSPenvironment(publiccloud)andstillexperienceInteltechnologybenefits.ThisreleasesupportsAmazonWebService(AWS)ElasticKubernetesService(EKS)andtheMicrosoftAzureKubernetes

Service(AKS)usingeitherHashiCorp*TerraformortheirnativeCLI.

TheBMRAandVMRAaretunedtosupportdiversedeploymentscenarios.Basedontheapplicationandnetworklocation,aBMRAorVMRAcanbeconfiguredforoptimaloperationandperformanceusingoneoftheprovidedConfigurationProfiles.

3

AnsibleplaybooksaredefinedbasedonthoseConfigurationProfiles,allowingfastandeasyauto-configurationandauto-

provisioningoftheReferenceSystemArchitectures.

3Workloadsandconfigurations.

Resultsmayvary.

UserManual|NetworkandEdgeReferenceSystemArchitecturesPortfolio

9

Figure4.

Figure5.

ContainerBareMetalReferenceSystemArchitecture(BMRA)

VMRA–VirtualClusteronaSingleNode

UserManual|NetworkandEdgeReferenceSystemArchitecturesPortfolio

10

Figure6.VMRA–VirtualClusteronMultipleNodes

Figure7.CloudRAforAmazonWebService(AWS)EKS

UserManual|NetworkandEdgeReferenceSystemArchitecturesPortfolio

11

Figure8CloudRAforMicrosoftAzureKubernetesService

2.2KeyElements

ThemainelementsoftheReferenceSystemArchitecturesincludethefollowing(referto

Figure9

foraBMRAexample):

.HardwareComponents:Multipleplatformhardwareoptionsareavailable,includingavarietyof4thand3rdGenIntel®

Xeon®ScalableprocessorSKUs,Intel®Xeon®DprocessorSKUs,Intel®EthernetNetworkAdapters,Intel®QAT,andIntel®ServerGPU.BIOSoptionsarelistedineachReferenceSystemArchitectureuserguide.DeploymentengineersshouldrefertotheappropriateuserguideduringdeploymenttoselectandconfigureoptimalBIOSvaluesbeforeclusterprovisioning.

Fordetails,see

Section3–HardwareComponents.

.SoftwareCapabilities:Thesoftwarecapabilitiesarebasedonopen-sourcesoftwaredeliveredbycloud-nativeandCNCFcommunitiesdrivingKubernetes,Istio,observability,DPDK,FD.io.OVS,OVS-DPDK,andthroughIntelGitHub.OptionsforRHELandUbuntuLinuxoperatingsystemsareavailable.ContainerenvironmentsuseDockercontainerruntimeaswellascontainerdandCRI-O.Fordetails,see

Section4–SoftwareComponents.

.ConfigurationProfiles:SpecifichardwareandsoftwareconfigurationsareprovidedintheConfigurationProfilesbasedonIntelassessmentandverification.Hardwareconfigurationsaddresstwoperformancecapabilities:“base”and“plus”.

.InstallationPlaybooks:Ansibleplaybooksimplementthebest-practiceconfigurationandsetupforeachConfiguration

Profiletoeaseandaccelerateinstallation.Formoredetailsaboutthehardwareandsoftwareconfigurationoptionsavailableandreadytouse,see

Section5–InstallationPlaybooks.

UserManual|NetworkandEdgeReferenceSystemArchitecturesPortfolio

12

Figure9.ReferenceArchitectureKeyElements–BMRAExample

2.3ConfigurationProfiles

2.3.1ConfigurationProfilesOverview

AConfigurationProfiledescribesaspecifichardwareandsoftwareBOMandconfigurations,applicableforaspecific

deployment

(Figure3)

.ConfigurationProfilestakeintoconsiderationthebest-knownconfiguration(BKC)validatedbyIntelforoptimizedperformance.

4

EachConfigurationProfileincludesinstallationscriptstodeploytherequiredcomponentsforaReferenceArchitectureFlavor.

InstallationscriptsareavailabletodeploytherequiredcomponentsforaReferenceArchitectureFlavor.EachReferenceArchitectureisbuiltonthefollowing:

.IntelPlatformfoundationwithIntelprocessorsandtechnologies.

.HardwareBOMoptimizedfordeliveringanapplicationataspecificlocationusingadeploymentmodel.Forexample,tosupportaUPFworkloadattheRemoteCO,theBMRAdeploymentispopulatedwiththemaximumavailablenetworkinterfacecards.

.SoftwareBOMleveragestheIntelplatformandenablescloud-nativeadoption.

.Installation(Ansible)PlaybookautomatestheinstallationofaReferenceArchitectureFlavorperaConfigurationProfilespecification.

ThisversionoftheNetworkandEdgeReferenceSystemArchitecturesPortfolioincludesthefollowingConfigurationProfilesthatsupportpernetworklocations.

.On-PremisesEdgeConfigurationProfile(CustomerPremises)

.AccessEdge(FarEdge)

.RemoteCentralOffice-ForwardingConfigurationProfile(NearEdge)

.RegionalDataCenterConfigurationProfile(RegionalPOP)

AdditionalConfigurationProfilessupportnon-location-specificdeployments:BasicandBuild-Your-Own.2.3.2On-PremisesEdgeConfigurationProfile(CustomerPremises)

ThisConfigurationProfileisdesignedforasmallclusterofstationaryormobileserverplatforms,rangingfromonetofour

servers.Usagescenariosincludedatacollectionfromsensors,local(edge)processing,andupstreamdatatransmission.Samplelocationsincludehospitals,factoryfloors,lawenforcement,media,cargotransportation,andpowerutilities.

ThisConfigurationProfilerecommendsaKubernetesclusterhardwareconfiguration,softwarecapabilities,andspecific

hardwareandsoftwareconfigurationsthattypicallysupportenterpriseedgeworkloadsusedinSmartCitydeployments,CDN,MediaAnalytics,andAd-insertion.ThisConfigurationProfileisdesignedtomeetthefollowing:

.Optimizeddataprocessing(togaininsightsandreduceupstreamtransmissionvolumes)ispriority

.Powerefficiencyisimportantinmostusecases

.Automatedinfrastructureorchestrationcapabilitiesaremandatory

.ComplexmediaanalyticspipelinesattheEdgearerequired

4Workloadsandconfigurations.

Resultsmayvary.

UserManual|NetworkandEdgeReferenceSystemArchitecturesPortfolio

13

2.3.3AccessEdgeConfigurationProfile(FarEdge)

TheAccessEdgeConfigurationProfilesetsanedgecomputinginfrastructureprimarilytargetingcellularaccessnetworks.ThisRAConfigurationProfileisdesignedforasmall,securedataprocessingclusterforhigh-speedandlowlatencyradioaccess

service.Thesenodesrepresentalargeportionofthetotalnumberofnetworkelementsforanoperatorwiththeequipment

possiblybeinginanoutsideplantinharsh,minimallycontrolledtemperaturecabinets,makingpowermanagementanecessity.ThereferencearchitectureusestheAccessEdgeconfigurationtosetuptheinfrastructureforvirtualizedradioaccessnetworks(vRAN)andFlexRAN™softwaresolutions.

ThisRAConfigurationProfileisdesignedtomeetthefollowing:

.Optimalbalanceofpowerandpacketperformanceforthegivenapplication

.Platformandnetworksecurity

.Lowlatencywithhigh-speedpacketprocessing

.Automatedinfrastructureorchestrationcapabilitiesaremandatory

2.3.4RemoteCentralOffice-ForwardingConfigurationProfile(NearEdge)

ThisConfigurationProfileisdesignedforclustersrangingfromahalfracktoafewracksofservers,typicallyinapre-existing,repurposed,unmannedstructure.Usagescenariosincl