1、CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSJake Saunders:Vice President,Asia-Pacific&Advisory Services Matthias Foo:Industry AnalystEXECUTIVE SUMMARYContainers are a foundational element for a telco network 5G Core(5GC)and for on-demand network slicing services,both of
2、which require light and scalable Network Functions(NFs).On the other hand,Virtual Machine(VM)technology is a well-proven and mature solution that has been implemented by many Communication Service Providers(CSPs)around the world and is expected to remain a key component of many networks in the comin
3、g years.As a result,CSPs are looking to telco cloud plat-forms to simplify and automate network operations to manage and orchestrate in-creasingly complex and disaggregated networks.However,the transition to autonomous telco cloud networks is not an easy feat,and industry players need to overcome va
4、rious challenges,including ensuring uniform data management,cloud technology fragmentation,and the need to maximize the value of previous network builds.To address these challenges,various innovations have been made.At the infrastructure and NF layers,technology is developing quickly to meet the nee
5、ds of the industry,including but not limited to:1)unified Physical Network Function(PNF),Virtual Network Function(VNF),and Cloud-Native Network Function(CNF)management;2)enabling consistent performance,reliability,and user experience in hybrid VNF/CNF environments;and 3)centralized support at the ha
6、rd-ware infrastructure layer.Whereas at the Management and Orchestration(MANO)lay-er,the autonomous Operations and Maintenance(O&M)and Lifecycle Management(LCM)process enabled by Artificial Intelligence Operations(AIOps)and continuous network monitoring,and analysis capabilities help operators move
7、away from manual operations toward intelligent and automated processes.TABLE OF CONTENTSExecutive Summary.1Industry Trends.2NFV Evolution Pain Points.5Cloud-Native Platform Developments.7Cloud Management and Orchestration Developments.8Future Evolution of Cloud Platforms.9Cloud Management(MANO)Evolu
8、tion.13Key Takeaways and Recommendations.14Glossary.152CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSThe industry is progressing at a rapid pace,with future and further enhancements of cloud-native platforms are already in the works.Hybrid VM/container deployments will eas
9、e the cloud-native trans-formation for CSPs,while unified hardware/software management and intent-based operations will drive networks toward increased automation.Higher levels of collaboration between Standards Devel-opment Organizations(SDOs)and open-source organizations will also enhance the inte
10、roperability of cloud-native technologies.In light of the above,ABI Research has provided some of our key takeaways,which are summarized as follows:CSPs need to keep up with industry changes or risk falling behind.Cloud-native platform providers need to meet and address the concerns of operators and
11、 help them transition seamlessly to cloud-native network architectures.SDOs and open-source organizations should work together to define common standards and APIs to ensure interoperability and the development of an open ecosystem.Figure 1:SWOT Analysis of Cloud-Native Architecture Migration(Source:
12、ABI Research)INDUSTRY TRENDSCloudification has been at the core of digital transformation for the telecommunications industry and is one of the key elements for transforming CSPs into Digital Service Providers(DSPs).EVOLUTION OF VNFS TO CNFSCloud-native networks will be the cornerstone for CSPs to u
13、nlock the commercial potential of a full 5G Standalone(SA)core network.CSPs across the world,such as Orange,AT&T,Vodafone,DISH,Ra-kuten Mobile,etc.,are investing in and have deployed cloud-native 5G SA core networks.The vision of cloud-native methodologies promotes scalability,complete resilience ac
14、ross hybrid(virtual and physi-cal)architectures,and ultra-rapid deployment and innovation cycles for new features,applications,and services.StrengthsImproved application performanceHigh application deployment flexibilityHigher efficiencyWeaknesses Decreased security across containersStandards are st
15、ill in development and evolvingRelative immaturity of ecosystem and solutionsOpportunitiesRapid development and support of telco cloud-native technologiesIncreased flexibility of cloud-native deployment facilitates migrationUnified hardware/software management and intent-based operations to drive in
16、creased network automationThreatsVarying scopes and requirements for data collection increase complexity of centralized data managementFragmentation of cloud technologies could lead to industry silosCSPs do not see the value of migration3CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS
17、&CONTAINERSOne key distinguishing feature of the evolution of VNFs to CNFs involves the use of containers.While VNFs abstract hardware to run Operating System(OS)instances,containers leverage a container en-gine to abstract the OS and divide a single OS instance into isolated environments that can r
18、un applica-tions.This allows for containerization,with software applications,functions,or microservices packaged with all the OS libraries and dependencies required to run code on the cloud.Another key difference is that CNFs built and deployed using cloud-native methodologies use what constitutes a
19、 key cloud-native pillarmicroservices.Microservices present an approach to software development that deconstructs the support and architecture of the applications portfolio into small,highly-cohesive,and loosely-coupled component-like services.By segmenting an application into mi-croservices,which c
20、an be reused for other applications,CSPs can create a more flexible and dynamic network infrastructure and processes.COEXISTENCE OF VMS AND CONTAINERSContainers are a foundational element for 5G and on-demand network slicing services,both of which require light and scalable NFs.The direct benefits o
21、f containers would be:Improved application performance,as containerized functions are simpler and stateless.Stateless programs are generally lighter(due to the absence of persistent local data)and easier to debug.The lightweight nature of containers allows for agile deployments of short-lived and ep
22、hem-eral servicesa distinct cloud-native feature that characterizes Continuous Integration/Continuous Delivery(CI/CD)networks.Deploymentflexibilityis enabled as a microservice architecture that allows applications to be developed independently without affecting the rest of the microservices that mak
23、e up the applica-tion.This is in contrast to the monolithic architecture of VMs,where the entire application needs to be scaled up to address component-level issues.RunningcontainersismoreenergyefficientthanVMs,as they do not require running a com-plete OS in each container,whereas each VM requires
24、spinning up its own OS.This would equate to less consumption of memory and CPU resources compared to VMs.On the other hand,VMs have been widely used by CSPs and have a more well-established solutions ecosystem.Some of the direct benefits of using VMs over containers would be:Increased security as VM
25、s run in isolation from other VMs on a shared host.This means that the compromise of an individual VM will not affect other neighboring VMs running on the same host.Well-definedNFVstandardsareestablished across various standardization bodies and open-source communities,ensuring interoperability of m
26、ultiple-vendor solutions.IncreasedmaturityofthetelcoNFVecosystemwith proven solutions supporting carrier-class capabilities such as high performance,availability,scalability,and security.Given the current benefits of both VNFs and CNFs,VM-container coexistence is expected to coexist for some time un
27、til the ecosystem for CNFs fully matures.4CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSTable 1:Comparison of VNFs and CNFs(Source:ABI Research)Virtual Network FunctionsCloud-Native Network FunctionsProsIncreased security across VMsWell-defined NFV standardsMaturity of eco
28、system and telco-grade solutionsImproved application performanceHigh application deployment flexibilityHigher efficiencyConsLower application performanceLower application deployment flexibilityDecreased security across containersStandards are still in development and evolvingRelative immaturity of e
29、cosystem and solutionsCSP REQUIREMENTSThe following section highlights some of the key requirements that CSPs may have as they migrate to cloud-native platforms:UNIFIED RESOURCE MANAGEMENTWith the emergence of multi-vendor and multi-technology environments,simplifying network opera-tions,resource ma
30、nagement,and orchestration via a unified single pane of glass Graphical User Inter-face(GUI)has become a key consideration for CSPs.Network resource management requirements are also expanding beyond single domain management(i.e.,core,transport,Radio Access Network(RAN),Multi-Access Edge Computing(ME
31、C)sites),to End-to-End(E2E)cross-domain orchestration.SEAMLESS AND FUTURE-PROOF MIGRATIONFor existing CSPs running live networks,ensuring that their subscribers are not impacted or affected during the migration process is a key priority.Cloud-native platforms and solutions need to consider how to co
32、ntinually ensure high network availability,reliability,and security throughout the entire mi-gration process to address such CSP concerns.Additionally,to future proof their solutions,cloud-native platforms need to be closely aligned with existing industry standards.PROTECTION OF EXISTING NETWORK INV
33、ESTMENTSGlobally,CSPs are all in different stages of their network evolution.While some CSPs have already moved to 5G SA networks,many CSPs are still in the midst of deploying their 4G networks.Particularly for CSPs who have just deployed or are still deploying 4G networks,moving to cloud-native net
34、works will be a challenge,given the large investments that they have just spent to roll out their current 4G networks.Cloud-native network solutions need to consider how they can help CSPs maximize their Returns on Investment(ROIs)and protect their existing investments.NETWORK AUTOMATIONAs CSPs begi
35、n to migrate to cloud-native networks,network automation is another key technology needed to address the increased complexities of managing disaggregated networks with the coex-istence of both VNFs and CNFs.Cloud-native networks also natively introduce increased digitization,in turn creating more bu
36、siness processes and interfaces.This native digital setting has a richer and more complex network processes,interfaces,and operations.In fact,it is reported that 400+network procedures need to be tracked in 5G,each with its own dedicated Key Performance Indicators(KPIs).That means that network param
37、eters,processing algorithms,and the number of alarms are expected to grow,increasing CSPs operational complexity.5CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSSUMMARY CONCLUSIONSThe shift to cloud-native networks inherently presents operators with an exciting opportunity
38、for growth due to its ability to support faster Time to Market(TTM)for network services with the use of microservices,CI/CD,and Continuous Testing(CT)pipelines.However,the coexistence of VNFs/CNFs and the emergence of multi-technology networks necessitate advanced capabilities of cloud-native platfo
39、rms.In this regard,ABI Research has identified unified resource management,seamless and future proof migrations,investment protection,and automation as key enabling factors to help CSPs manage these increasingly disaggregated and complex networks.NFV EVOLUTION PAIN POINTSThis section explores some o
40、f the challenges faced by CSPs as they migrate from physical,to VM,and to cloud-native network architectures.Figure 2:Key Pain Points of NFV Evolution(Source:ABI Research)LIMITATIONS IN NFV AUTOMATIONA major hurdle that CSPs need to overcome in order to successfully deploy automation in the network
41、is related to data collection and management.Accurate and real-time data are a necessary component for closed-loop network automation and feedback.However,the multi-vendor architecture of cloud-native networks poses challenges to data management in the following areas:Scope of Data Collection:At pre
42、sent,third-party data collection and Network Data Analytics Function(NWDAF)solution vendors mainly focus on the collection of performance and fault events from the virtualized layer(i.e.,the VM and containers).However,there is a lack of information cap-tured on the underlying hardware performance an
43、d status.Requirements on Data/Collection Usage:Multiple tenants on the same cloud-native platform may have different requirements and policies on data management(e.g.,collection,formatting,usage,etc.),thereby leading to increased complexities of implementing centralized data analytics functions.Data
44、 ManagementCloud Technology FragmentationCarrier-Class RequirementsLimited scope of data collectionDiffering tenant data policies Standards Development Organizations Open-source communities and projectsChallenges in achieving high network availability requirement of 99.999%Data security and sovereig
45、nty needs6CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSCLOUD TECHNOLOGY FRAGMENTATIONWith the move to cloud-native networks,the industry is moving away from traditional methods of de-fining international telecommunication standards via SDOs,to open-source communities and
46、projects,as commonly practiced by the IT community.However,this has led to the fragmentation of the telco cloud ecosystem with varying standards and open-source solutions.Additionally,with new build public clouds also bringing their own cloud stacks to the market,the above factors could potentially
47、lead to increasingly siloed operating environments if interoperability and seamless migration between solu-tions are not supported.Some examples of different standards and open-source solutions are listed below.ETSIThe European Telecommunications Standards Institute(ETSI)is a traditional SDO that in
48、troduced the NFV concept back in 2012 and has released a number of NFV specifications in the form of NFV Re-leases 1 to 4.However,as telco networks started to move from NFV to cloud-native architectures,the SDO has evolved its existing NFV-MANO standards to address containerized technologies as well
49、,with functions,such as Container Infrastructure Services Management and Center for Internet Security Cluster Management,added in NFV Release 4.NEPHIONephio is an open-source Linux Foundation project that aims to provide carrier-grade,simple,open,Kubernetes(K8s)-based cloud-native intent automation
50、and common automation templates that sim-plify the deployment and management of multi-vendor cloud infrastructure and NFs across extensive edge deployments.In August 2023,Nephio R1 was released and introduced features to enable CSPs to efficiently manage cloud infrastructure,NFs,K8s clusters,and net
51、work configurations.SYLVASylva is an open-source community project hosted by the Linux Foundation Europe.Announced in November 2022,this project aims to develop a cloud-native infrastructure stack to host telco and edge use cases.Key challenges that Sylva targets to address include network performan
52、ce,distributed cloud(K8s cluster LCM,hybrid deployments of containers on OpenStack/VMWare or on bare metal),energy efficiency,security(hardening and compliance),and openness with standardized Application Programming Interfaces(APIs).AVAILABILITY AND SECURITY OF TELCO CLOUD PLATFORMS For CSPs,maintai
53、ning high network availability of 99.999%or better is a key requirement as any net-work outage can lead to severe consequences,such as disruptions to services,loss of revenues,and even penalties.The evolution to cloud-native architectures require platforms to provide carrier-class high availability
54、and reliability,which can be a challenge for solution providers.In addition,CSPs also need to be cognizant of other regulatory requirements such as ensuring data security and sovereignty.SUMMARY CONCLUSIONSThe telco cloud-native industry is faced with a number of challenges,including ensuring effici
55、ent data management,potential cloud technology fragmentation,and the need to meet the high availability and security requirements for telco cloud networks.These are key issues that next-generation cloud-native platforms should target to address.7CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR S
56、TANDARDS&CONTAINERSCLOUD-NATIVE PLATFORM DEVELOPMENTSThis section explores the latest developments of cloud-native platform technologies.Figure 3:Example of Cloud-Native Platform Architecture(Source:ABI Research)IMPLEMENTING A VERSATILE CLOUD-NATIVE PLATFORMAs mentioned above,to support the coexiste
57、nce of VNFs and CNFs,multiple cloud-native platform solution providers are offering network orchestration solutions that support both VM-and container-based NFs concurrently.Additionally,various cloud-native platforms are also offering CSPs the flex-ibility to support VMs,containers,and bare metal s
58、ervers under a telco cloud platform.For example,VMWares Workflow Hub for Telco Cloud Automation provides CSPs with a single platform covering an E2E automation solution for bare metal,VMs,and containers.Huaweis Telco Cloud platform and E2E automation solutions offer increased flexibility,supporting
59、VMs,VM-based containers,and bare metal-based containers on a single platform,while Nokias Cloud Packet Core solution supports the unified management of VNFs,CNFs,and PNFs.DELIVERING A CONSISTENT VM AND CONTAINER EXPERIENCERather than simply being able to run VNFs and CNFs in conjunction with one ano
60、ther,cloud-native platforms also support resource pooling and allocation capabilities,enabling network resources to be allocated based on specific NFs requirements.This capability enables consistent performance,reliabil-ity,and user experience in a hybrid VNF/CNF environment.UNIFIED HARDWARE MANAGEM
61、ENTIn an open cloud-native architecture,in addition to supporting multi-vendor VNFs and CNFs,multi-vendor support is also required at the physical infrastructure layer where heterogeneous hardware resources,such as x86,Advanced RISC Machine(ARM)processors,Graphics Processing Units(GPUs),Data Process
62、ing Units(DPUs),and Smart Network Interface Cards(SmartNICs),will be used in net-works to meet different use cases.Physical ComputePhysical StoragePIMHardware LayerVIM/CIMVirtualization LayerPhysical NetworkVirtual ComputeVirtual StorageVirtual NetworkNetwork Function Virtualization InfrastructureOp
63、enStackKubernetesVNFM/CNFMNetwork FunctionsOSS/BSSNFVONFV MANOBare MetalCloud PlatformsVNFVNFVNFCNFCNFCNF8CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSTo meet this challenge,Physical Infrastructure Manager(PIM)solutions are being offered in the market to enable centralize
64、d management of hardware devices in the resource pool,providing features such as configuration management,fault management,and performance monitoring.PIMs generally also support open API northbound interfaces to connect to upper layers,such as the MANO and Opera-tions Support System(OSS)/Business Su
65、pport System(BSS)layers,to simplify and automate E2E ap-plication development,deployment,and resource management.SUMMARY CONCLUSIONSCloud-native platform technology is developing quickly to meet the needs of the industry,including but not limited to:1)unified PNF,VNF,and CNF management;2)enabling co
66、nsistent performance,reliabil-ity,and user experience in hybrid VNF/CNF environments;and 3)centralized support at the hardware infrastructure layer.CLOUD MANAGEMENT AND ORCHESTRATION DEVELOPMENTSThis section explores the latest developments of cloud MANO technologies.IMPLEMENTING AUTOMATION AND FACI
67、LITATING OPERATIONS&MAINTENANCETo address the automation requirements of CSPs,there are a variety of MANO solutions in the market that support various automated O&M use cases,such as:Analysis and filtering of network alarms and alerts.Expedited fault identification and recovery.Improving the monitor
68、ing and management of end user service experience.Providing network assurance in accordance with Service-Level Agreements(SLAs).Many cloud-native platforms are also integrating Artificial Intelligence(AI)and Machine Learning(ML)capabilities into their solutions to enable AIOps.AIOps appears to be ga
69、ining traction,particularly in the area of O&M,with key objectives being increased efficiency,reliability,and network performance.In particular,the development of generative AI solutions appears to be gaining traction in the industry,with many cloud-native platform providers looking to incorporate t
70、his technology into their solutions.At the moment,generative AI solutions appear to be focused on improving operational efficiencies,supporting features such as intelligent assistance,fault identification and rectification,and customer self-help services via the use of Large Language Models(LLMs).LI
71、FECYCLE MANAGEMENT OF NETWORK FUNCTIONSCloud-native platforms are also introducing automation into VNF/CNF/cloud infrastructure LCM(i.e.,onboarding,instantiation,termination,upgrade,admin scaling,and NFs support)by supporting De-vOps and CI/CD/CT pipelines,thereby enabling CSPs to quickly deploy con
72、figuration changes and de-crease the TTM for new services.Various cloud-native platforms also support CNF and Container-as-a-Service(CaaS)LCM capabilities across CaaS vendors,thereby increasing deployment flexibility for multi-vendor environments.With a simple-to-use GUI interface,LCM solutions help
73、s CSPs reduce manual errors during the config-uration process,proactively monitor KPIs during change operations and supporting one-click rollbacks in the event of errors.When integrated with AIOps,CSPs are able to analyze network performance after deployment of the code,thereby enabling closed-loop
74、operations.9CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSSOFTWARE-DEFINED NETWORKINGSoftware-Defined Networking(SDN)is a networking approach where software-based controllers are used to communicate with hardware infrastructure to direct network traffic via the use of APIs
75、.Some of the key benefits of implementing SDN are detailed below:DynamicNetworkControl:As opposed to the traditional method of using dedicated hardware devices(i.e.,routers and switches)to control network traffic,the software-based SDN approach accords CSPs with more flexibility and allows them to c
76、ontrol the network,change configuration settings,provision resources,and increase network capacity from a centralized user interface in real time.IncreasedDeploymentFlexibility:SDN allows CSPs to use a single protocol to communicate with heterogeneous hardware devices through a central controller,th
77、ereby not only simplifying network design and operations,but also giving CSPs the flexibility to select networking equipment from multiple vendors.HigherNetworkSecurity:SDN provides CSPs with increased visibility over the entire network and enables network administrators to set access control and se
78、curity policies across the network from one central location.It also provides CSPs with the ability to create separate security zones for devices based on their security requirements,or immediately isolate compromised equipment to contain the impact to the network.CROSS-DOMAIN ORCHESTRATIONWith an i
79、ncreasingly disaggregated and distributed network spanning multiple domains such as the core,RAN,transport,and edge,cross-domain orchestration solutions have emerged in the industry.These solutions enable operators to not only manage the E2E delivery and management of network slicing services,but to
80、 also provide network assurance to their customers.SUMMARY CONCLUSIONSCloud MANO technologies are critical components needed to manage network O&M,NF and services LCM,and traffic.These developments are key in helping operators move away from manual opera-tions toward intelligent and automated proces
81、ses.With continuous network monitoring and analysis capabilities,cloud MANO technologies help CSPs maximize network efficiency,while reducing opera-tion downtimes.FUTURE EVOLUTION OF CLOUD PLATFORMSThis section explores the potential of future evolutions of cloud-native platform technologies.BACK-TO
82、-BACK DEPLOYMENTSTo prevent any disruptions to existing services running on VMs,many CSPs have chosen to deploy their cloud-native networks as new build deployments alongside existing virtualized infrastructure,in what ABI Research terms as a“back-to-back”deployment model.In such a model,separate in
83、frastructure is used for VNF and CNF deployments,with centralized resource management and LCM performed at the orchestration layer to avoid impacting services on live networks.10CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSHowever,as this model requires new infrastructure
84、,it can result in higher deployment costs.There-fore,the“back-to-back”deployment model may be better suited for new licensees and networks,as opposed to existing CSPs with well-established networks.Figure 4:Typical Architecture for a Telco Cloud Data Center Network(Source:ABI Research)BARE-METAL CON
85、TAINERS VERSUS VM-BASED CONTAINERSThere has been a significant debate between the use of bare-metal and VMs for container deploy-ments.On one hand,bare-metal servers are dedicated servers that have better compute capabilities compared to VMs due to:1)their direct access to the processing resources;a
86、nd 2)how they operate with less compute overhead from the lack of a hypervisor.On the other hand,VM-based containers offer a wide range of advantages over bare-metal containers,including:1)better resource utilization in multi-tenant environments;2)higher degree of decoupling between CNFs and the inf
87、rastructure layer;and 3)better security due to the hypervisors strict resource isolation policies.Containers will be gradually relied upon more as CSPs transition toward cloud-native networks and the selection of the right container deployment model will be highly dependent on where CSPs are in this
88、 transition.In the interim,ABI Research expects initial CSP cloud-native deployments to be oriented more toward building their stack on their existing virtualization layers,relegating the bare-metal de-ployment option to the secondary option for now until the technology and ecosystem become more mat
89、ure.InternetFirewallFirewallCore SwitchesCore SwitchesAggregation SwitchesAggregation SwitchesTOR SwitchesServersServersControllerHard DiskStorage PlaneManagement PlaneService Plane11CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSTable 2:Comparison between Bare-Metal Contai
90、ners and VM-Based Containers(Source:ABI Research)CriteriaVM-Based ContainersBare-Metal ContainersDeployment Time When the VM is not ready:minute-level+creation timeSeconds-level spin-upResource ManagementUnified resource management across multi-tenant environments Resource pools are divided into sil
91、o modes,reducing resource usage efficiency in multi-tenant environmentsSecurityHigh security due to the hypervisors strict resource isolation policiesLack of resource isolation of bare metals(due to a single host OS)and shared kernels result in a less secure environmentO&MHigh degree of decoupling b
92、etween CNFs and the infrastructure layerCNFs and the infrastructure layer are tightly coupled,thus complicating operations and maintenance processes(e.g.,CNF upgrades,etc.)Ecosystem MaturityHighly mature ecosystemRelatively immature ecosystemSUPPORTING SMOOTH EVOLUTION TO CLOUD-NATIVE NETWORKSTo sup
93、port a smooth transition to cloud-native networks,ABI Research has identified two key enabling factors,as discussed below.HYBRID VM/CONTAINER DEPLOYMENT MODELNew solutions are emerging where both VMs and containers can be supported by a single unified infrastructure,termed as“hybrid VM/container”dep
94、loyments,which provide a range of benefits,including:Seamless network evolution and upgrades while ensuring high network availability Centralized resource management with single cloud deployments Unified cloud security and assurance Investment protection with the reuse of existing infrastructureTo i
95、llustrate the potential savings enabled by such hybrid VM/container deployments,ABI Research has done a comparison between the Capital Expenditure(CAPEX)required for back-to-back new builds against hybrid VM/container deployments.Based on a scenario where there is a deployment of 32 computing nodes
96、for 5GC,the hybrid VM/container deployment model is expected to save operators up to 32%of CAPEX compared to a back-to-back new build model.These savings mainly result from the reuse of infrastructure equipment such as core and aggregation switches,firewalls,SDN con-trollers,storage disk array contr
97、ollers,and management plane servers.Additionally,manpower costs needed to redesign the core network architecture are also reduced.XXXXX4444412CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSTable 3:CAPEX Comparison between Back-to-Back New Build and Hybrid VM/Container Deplo
98、yment Models;Scenario of Network Deployment/Expansion of 32 5GC Computing Nodes(Source:ABI Research)Cost ItemTypical List Price(US$000s)Back-to-Back New BuildHybrid VM/ContainerQuantityTotal (US$000s)QuantityTotal (US$000s)NetworkService Plane Top-of-Rack(TOR)Switches2761626162Storage/Manage-ment Pl
99、ane TOR Switches1961146114Aggregation Switches(End of Row(EOR)56211200Core Switches (Data Center-Gateway(DC-GW)100220000Firewall1022000SDN Controller150230000StorageStorage Disk Array Controller4328600Hard Disks11222222ComputingManagement Plane Servers353105135Service Plane Servers40321,280321,280To
100、tal Equipment Cost2,4011,613Core Network Manpower Installation Costs8/month3242.520Data Network Manpower Installation Costs8/month0.2520.252Total CAPEX2,4271,635ETSI STANDARDS EVOLUTIONAs referenced in an earlier section,fragmentation in cloud technologies and standards could potential-ly lead to in
101、creasingly siloed solutions.To enable a smooth transition to and interoperability of cloud solutions,SDOs such as ETSI and open-source organizations should continue to work closely together to leverage the advantages that both types of organizations can bring to the table.In this regard,ETSI has ann
102、ounced plans for the next decade of NFV development and is currently preparing NFV Release 6,with key action items that include:Enhancing and evolving NFV-MANO APIs,including the introduction of“declarative”and“intent-driven”APIs,as well as increased alignment with open-source de facto standard APIs
103、.Improving the general capabilities for NFV-MANO with capabilities such as increased automation,supporting greater distributed deployments,and predictive maintenance.Supporting new virtualization technologies beyond traditional VMs and containers.Integrating heterogenous hardware,APIs,and cloud plat
104、forms through a unified management framework.And increasing alignment with other industry organizations,including open-source communities.13CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSSUMMARY CONCLUSIONSIn terms of telco cloud platforms,one of the main themes that stood
105、out was the importance placed on simplifying deployments of and migrations to cloud-native networks.This was clearly demonstrated with the following:1)the industry focusing on developing centralized and unified multi-technology management platforms;2)ensuring investment protection from previous buil
106、ds to lower Total Cost of Ownership(TCO)considerations;and 3)moves to ensure standards alignment across SDOs and open-source organizations.CLOUD MANAGEMENT(MANO)EVOLUTIONThis section explores the potential of future evolutions of cloud-native platform technologies.MANAGEMENT PROCESSESFuture NFV fram
107、eworks are envisioned to provide unified management for heterogeneous hardware and virtualized infrastructure,thereby providing operators with E2E Fault,Configuration,Accounting,Performance and Security(FCAPS)monitoring and visualization capabilities across multi-vendor multi-component environments.
108、NFV standards are also expected to improve the consistency of data man-agement across solutions,thereby enabling uniformed collection and analysis of data across complex networks.Advanced O&M capabilities,driven by the integration of AI/ML,will also enable CSPs to implement higher-level applications
109、 across various use cases such as disaster recovery,redundancy design,pre-dictive maintenance,digital twins,precise fault isolation,resource pool management,etc.,thereby helping operators achieve High Availability(HA)(i.e.,five-nines availability or higher)networks.INTENT-DRIVEN AUTOMATIONTo address
110、 the increasingly complex network environment,there is also a strong push toward generic management solutions such as declarative,intent-based operations.For intent-based deployment models,an operator will only need to specify their objectives based on a small number of parameters.Thereafter,declara
111、tive,intent-driven APIs will translate these intents into a format understood by the NFV-MANO system,which will,in turn,manage the lifecycles for VNFs and CNFs in accordance with the desired outcome.As a result,declarative,intent-based operations are expected to greatly simplify and automate the net
112、work management and configuration process,as well as minimize potential errors that could be made during manual configuration operations.Generative AI is also expected to play a role in this space.As technology evolves,many cloud-native solution providers are expecting generative AI to help boost th
113、e revenue of operators by assisting in software code generation and configuration,thereby enabling faster TTM of new services.Generative AI can also be potentially integrated with declarative intent-based operations to further simplify the network configuration process.SUMMARY CONCLUSIONSCloud MANO
114、solutions are expected to evolve to further enhance network O&M processes,support-ing higher levels of network automation with increased centralized management and visibility into the network,declarative,intent-based operations,integrated advanced AI/ML capabilities,and new service offerings.These d
115、evelopments are expected to greatly benefit operators by not only reducing Operational Expenditures(OPEX)for network operations,but also drive revenue by reducing the TTM for new services.14CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSKEY TAKEAWAYS AND RECOMMENDATIONSThro
116、ugh studying the development and evolution of cloud-native platforms,ABI Research has distilled the key takeaways outlined below.COMMUNICATION SERVICE PROVIDERSKeepingUpwithIndustryChanges:With the shift to cloud-native environments,CSPs need to move away from traditional ways of working and embrace
117、 new technologies or risk falling behind.CSPs that are not seriously considering a migration to cloud-native networks could miss out on the benefits provided by the use of containers,such as increased scalability and faster TTM for new services.To address CSPs concerns,the latest developments in tel
118、co cloud technologies support seamless migrations and protect existing network investments with hybrid VNF and CNF deploy-ments with high reliability and security.CLOUD-NATIVE PLATFORM PROVIDERSMeetingtheConcernsofOperators:Protecting their existing investments in VM technolo-gies will likely contin
119、ue to be a key priority for CSPs and this should be addressed by cloud-native platform providers.Solutions need to help CSPs transition seamlessly to cloud-native network ar-chitectures with unified management and orchestration capabilitiescovering VMs,containers,hardware,and multi-generational tech
120、nologiessupport carrier-grade high reliability and provide future proof solutions that comply with industry standards.THE INDUSTRY IN GENERAL Industry Consensus:The current fragmentation of cloud technologies is a key concern,as it could potentially lead to increasingly siloed solutions.To ensure in
121、teroperability and the development of an open ecosystem,SDOs and open-source organizations should work together to define common standards and APIs.This will ensure that CSPs can seamlessly integrate solutions across vendors to create“best-of-breed”networks.15CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS
122、:OUTLOOK FOR STANDARDS&CONTAINERSGLOSSARYAIArtificialIntelligence:The ability for a computer(physical or digital)to perform tasks commonly associated with a human or intelligent being.AIOps AI Operations:The use of AI to automated information technology operations.BSS Business Support System:A syste
123、m used to support business and/or customer-related functionalities.CaaS Container-as-a-Service:A cloud service that helps customers manage containers.CAPEXCapitalExpenditure:Resources used to purchase assets.CI/CD Continuous Integration and Continuous Delivery:A process where code changes are made t
124、o a codebase simultaneously and automatically.CNFCloud-NativeNetworkFunction:Software service that performs network functions based on cloud-native principles.CPU Central Processing Unit:A specialized processor designed to perform core computing capabilities.CSP Communication Service Provider:An ent
125、erprise that focuses on offering connectivity services to its subscribers.DPU Data Processing Unit:A specialized processor designed to handle data-related workloads.DSP Digital Service Provider:A company that provides digital services.E2E End-to-End:A full process from beginning to end.FCAPSFault,Co
126、nfiguration,Accounting,PerformanceandSecurity:A framework for network management.Gen AI Generative AI:Artificial intelligence capable of generating text,images,or other types of media.GPUGraphicsProcessingUnit:A specialized processor designed to accelerate the rendering of graphics.GUIGraphicalUserI
127、nterface:Digital interface that allows users to interact with graphical elements such as icons and buttons.KPI Key Performance Indicator:Measurement of performance against a particular quantifiable objective.K8sKubernetes:An open-source system for deployment,scaling,and management of containers.LCML
128、ifecycleManagement:Managing the lifecycle of a network function or service,such as onboarding,instantiation,termination,upgrade,etc.LLMLargeLanguageModel:Models trained using large datasets.MANOManagementandOrchestration:Framework for managing and orchestrating networking functions.MECMulti-AccessEd
129、geComputing:Moves computing functions from a centralized cloud to the edge of the network and closer to the customer;the network edge then analyses,processes,and stores the data.16CLOUD-NATIVE PLATFORMS&MANAGEMENT TRENDS:OUTLOOK FOR STANDARDS&CONTAINERSMLMachineLearning:A technique that allows compu
130、ters to learn from studying and analyzing data.NFNetworkFunction:A component of network infrastructure that serves a defined purpose.NFVNetworkFunctionVirtualization:A network architecture that virtualizes network services that have traditionally been run on proprietary hardware.NWDAFNetworkDataAnal
131、yticsFunction:A network function for collecting and analyzing data from different systems within the 5G network.OPEXOperationalExpenditure:Expenses required to maintain work operations.OS Operating System:A system that manages all of the other applications and resources.OSS Operations Support Servic
132、e:A system used to support network operation-related functionalities.O&MOperations&Maintenance:A combination of management,monitoring,and upkeep of a facility to ensure proper functioning of equipment.PIMPhysicalInfrastructureManager:A system to manage the physical infrastructure of the network.PNFP
133、hysicalNetworkFunction:The use of proprietary hardware to perform defined network functions.RANRadioAccessNetwork:Connects end-user devices to the core network via radio links.RCA Root Cause Analysis:A method to identify the underlying cause of the issue or problem.ROI Return on Investment:A measure
134、 of profitability by comparing the final value generated by a product against its initial cost.SA Standalone:5G network with dedicated equipment and network functions,5G radios coupled with cloud-native core network,network functions completely virtualized in the cloud.SDNSoftware-DefinedNetworking:
135、A networking approach where software-based controllers are used to communicate with hardware infrastructure to direct network traffic via the use of APIs.SDO Standards Development Organization:An organization that aims to develop technical standards to meet the needs of an industry.SmartNICSmartNetw
136、orkInterfaceCard:A programmable accelerator that can handle network-related tasks.TTMTimetoMarket:Time taken to bring a product from idea conception to actualization.VMVirtualMachine:A compute resource that uses software instead of a physical equipment to run programs and applications.VNFVirtualNetw
137、orkFunction:Virtualization of network functions previously performed via the use of proprietary hardware.Published November 2023ABI Research157 Columbus AvenueNew York,NY 10023 United StatesTel:+1 516-624-About ABI ResearchABI Research is a global technology intelligence firm delivering actionable r
138、esearch and strategic guidance to technology leaders,innovators,and decision makers around the world.Our research focuses on the transformative technologies that are dramatically reshaping industries,economies,and workforces today.ABI Researchs global team of analysts publish groundbreaking studies
139、often years ahead of other technology advisory firms,empowering our clients to stay ahead of their markets and their competitors.2023 ABI Research.Used by permission.ABI Research is an independent producer of market analysis and insight and this ABI Research product is the result of objective resear
140、ch by ABI Research staff at the time of data collection.The opinions of ABI Research or its analysts on any subject are continually revised based on the most current data available.The information contained herein has been obtained from sources believed to be reliable.ABI Research disclaims all warranties,express or implied,with respect to this research,including any warranties of merchantability or fitness for a particular purpose.