《易观梅森:构建具有成本效益的云原生5G SA网络:与TCO对比(2023)(英文版)(27页).pdf》由会员分享,可在线阅读,更多相关《易观梅森:构建具有成本效益的云原生5G SA网络:与TCO对比(2023)(英文版)(27页).pdf(27页珍藏版)》请在三个皮匠报告上搜索。
1、 BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISONFEBRUARY 2023 BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISONGorkem Yigit,Caroline Chappell and Luwen MengPERSPECTIVE1.Executive summary 31.1 The TCO of on-premises Azure Operator Nexus deployments is 38%lower
2、than that for DIY deployments 31.2 Operators can further optimize cloud-native network TCO by using Azure Operator Nexus hybrid cloud architecture 52 5G networks are driving operator investment in cloud-native infrastructure 62.1 The mobile core is responsible for the majority of operators spending
3、on network clouds 62.2 Cloud-native infrastructure and automation are key to unlocking the benefits of 5G 73.Cloud-native networks are driving a shift from vertically integrated stacks to disaggregated cloud platforms 83.1 Operators must decide how to source cloud-native infrastructure for 5G 83.2 V
4、endor-integrated stacks limit flexibility and choice,but retain appeal for certain operators 83.3 DIY private clouds require deep pockets to build and maintain 93.4 PCPs offer a new approach to network clouds based on their hyperscale cloud technologies and high levels of 9 automation3.5 Operators m
5、ust understand the full cost implications of disaggregated deployment models to guide their cloud 10 platform choices4.Azure Operator Nexus can reduce the TCO of 5G SA network cloud infrastructure by up to 43%compared to a DIY private cloud 114.1 The hybrid Azure Operator Nexus model offers the lowe
6、st overall TCO of all scenarios 134.2 The cloud infrastructure and operations TCO of the Azure Operator Nexus on-remises scenario is up to 38%l ower than that for the DIY model due to a reduction in opex 144.3 Adopting a hybrid cloud architecture based on Azure Operator Nexus could result in a cloud
7、 infrastructure and operations TCO reduction of 43%165.Conclusions and recommendations 216.Annex 236.1 Connections and traffic 236.2 Cost parameters 236.3 TCO breakdown for Tier-1 operator in Latin America for all three deployment scenarios 257.About the author 26Contents BUILDING COST-EFFICIENT CLO
8、UD-NATIVE 5G SA NETWORKS:A TCO COMPARISON3The 5G network has been conceived as a cloud-based network in which all functions are expected to run on a horizontal network cloud rather than on siloed virtualized platforms.Leading telecoms operators are beginning to deploy the 5G standalone(SA)core;this
9、is a set of cloud-native network functions(CNFs)that will require operators to accelerate their adoption of cloud-native infrastructure.The cloud-native 5G SA core also requires operators to commit to a cloud-first,horizontal approach to running the network with very high levels of automation.Cloud-
10、native clouds enable operators to adopt open and disaggregated archi-tecture;cloud-native infrastructure is sourced independently from vendors CNFs.However,operators must choose between two main approaches to disag-gregated cloud deployment.The first option is to build a private network cloud by int
11、egrating multiple technology components including compute and storage hardware,cloud technology platforms,software-defined networking(SDN)and,in advanced cases,orches-tration capabilities.We call this the do-it-yourself(DIY)private cloud deployment model.The alternative is to use the hyperscale-base
12、d infrastructure and advanced automation frameworks offered by public cloud providers(PCPs).Operators that are planning their 5G SA core deployments and are looking to deploy virtualized/Open RANs in the future must decide on their approach to cloud-native infrastructure;a holistic total cost of own
13、ership(TCO)analysis is critical.Analysys Mason,in collaboration with Microsoft,has analyzed the TCO of deploying a cloud-native 5G SA network using a PCP cloud solution based on Azure Operator Nexus compared with that using the DIY private cloud model.The TCO model considers three green-field deploy
14、ment scenarios for the cloud-native 5G SA core in consumer macro networks,modeled for two hypothetical Tier-1 operator profiles(in Western Europe and in Latin America).In this report,we discuss the key findings of this analysis(Figure 1.1).1.1 The TCO of on-premises Azure Operator Nexus deployments
15、is 38%lower than that for DIY deployments Our 5-year TCO model shows that the Azure Operator Nexus on-premises scenario has a TCO for cloud-native 5G SA deployments that is up to 25%lower than that for the DIY private cloud model.The Azure Operator Nexus on-premises scenario can also offer a TCO red
16、uction of up to 38%for cloud infrastructure(CaaS,PaaS,OS,SDN and cloud hardware)and operations,when 1 Executive summary BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON4excluding CNF-related capex and opex components,which are assumed to be equal for all scenarios.The main driver
17、 of the TCO savings in the Azure Operator Nexus scenarios is the reduction in opex that is enabled by the more automated and efficient cloud infrastructure and CNF operations that are supported by Azure Operator Nexuss comprehensive cloud platform and as-a-service operations model.Indeed,our TCO stu
18、dy revealed that the opex associated with building,deploying and maintaining a DIY-based cloud-native network can be substantial,and operators often do not have a complete understanding of these costs.Our research and interviews with large Tier-1 operators indicate that operators face major challeng
19、es with managing their private clouds and on-premises data centers,which leads to high operating expenses in terms of staffing and tools.Our model shows that Azure Operator Nexuss comprehensive and consistent cloud environment,automation framework and managed service capabilities can reduce the numb
20、er of repetitive,time-and resource-con-suming and error-prone tasks for the main cloud-native network operations by 58%,resulting in opex savings of up to 36%compared to DIY private cloud model.The details of these operational efficiencies are discussed in section 4.2.The use of the Azure Operator N
21、exus cloud platform can also lead to a reduction in capex when compared to a DIY private cloud model.Using Azure Operator Nexuss predesigned and FIGURE 1.1:OVERVIEW OF THE TCO ANALYSIS RESULTS SOURCE:ANALYSYS MASON,202325%28%38%43%44%Cloud infrastructure and operations TCO savings compared to the DI
22、Y modelOpex savings compared to the DIY modelOverall TCO savings compared to the DIY modelAzure Operator Nexus on-premises(5-year)Azure Operator Nexus hybrid cloud(5-year)BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON5validated infrastructure greatly reduces the costs associate
23、d with implementing and integrating cloud hardware as well as testing.This results in a capex saving of 1117%.1.2 Operators can further optimize cloud-native network TCO by using Azure Operator Nexus hybrid cloud architectureOur TCO analysis shows that deploying a cloud-native 5G SA network using Az
24、ure Operator Nexus hybrid cloud architecture can result in even greater cost savings over a DIY private cloud model.Indeed,the overall TCO can be reduced by an additional 2%and 3%for a Western European operator and a Latin American operator,respectively,by deploying a set of control plane functions
25、and management applications on Azure public cloud,while the remaining network functions are on-premises in operator data centers.These results are discussed in section 4.3.These further TCO savings are primarily attributed to a reduction in on-premises hardware costs,a shift from high up-front capex
26、 to a more optimized consumption-based opex and the ability to avoid overprovisioning hardware for peak usage.These savings can poten-tially be further optimized by better workload allocation in the public cloud environment.BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON62 5G ne
27、tworks are driving operator investment in cloud-native infrastructure 2.1 The mobile core is responsible for the majority of operators spending on network clouds The 5G network has been conceived as a cloud-based network in which all functions are expected to run on a horizontal network cloud rather
28、 than on siloed virtualized platforms.It is therefore not surprising that mobile core deployments are the main driver of operators network cloud investments(Figure 2.1).Many operators started their journey to the network cloud with the 4G evolved packet core(EPC),which is a key constituent of the 5G
29、 non-standalone(NSA)core.Leading operators are now beginning to deploy the 5G SA core;the pace of these deployments will increase from 2023 and will drive spending growth on network cloud infrastructure for the mobile core over the next few years.In particular,investing in the 5G SA core will requir
30、e operators to accelerate their adoption of cloud-native infrastructure,which will need to be operated and managed using cloud-native automation.Network function virtualization(NFV)was conceived,as far as possible,as an extension of operators existing opera-tions,while the cloud-native 5G SA core re
31、quires operators to commit to a cloud-first approach to running the network.Experience with deploying and automating the cloud-native mobile core will serve operators well as they move to applying cloud-native technologies to the RAN in the future.FIGURE 2.1:OPERATORS SPENDING ON NETWORK CLOUD INFRA
32、STRUCTURE FOR THE MOBILE CORE,WORLDWIDE,20212027 SOURCE:ANALYSYS MASON,2023024681012Spending(USD billion)20224202520262027 BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON7FIGURE 2.2:DRIVERS OF IMPLEMENTING A 5G SA CORE,WORLDWIDE,OCTOBER 20221 SOURCE:ANALYSYS MASON,202
33、32.2 Cloud-native infrastructure and automation are key to unlocking the benefits of 5GThe 5G SA core is key to operators ambitions to generate new enterprise revenue growth using advanced services such as network slicing.Operators need specific technical capabilities to achieve network slicing,incl
34、uding the ability to scale cloud-native mobile core functions up and down to meet the needs of individual customers and to orchestrate functions end-to-end to create customized private networks.A recent Analysys Mason survey confirms that such capabilities are some of the top benefits that operators
35、 expect from the 5G core,alongside the ability to both differentiate and innovate services in supporting roles and achieve full automation(Figure 2.2).The benefits of scalability,flexibility and orchestration are intrinsically delivered by cloud-native clouds that have full cloud-native automation c
36、apabilities.Operators are finding that their choice of cloud-native network cloud can signifi-cantly influence the performance and cost of a mobile core based on the level of cloud-native automation that such an environment can provide.1 What are the most important drivers for implementing a 5G SA c
37、ore,in terms of your commercial model?;n=68.0%10%20%30%40%50%60%Percentage of operators placing in their top threeNetwork differentiation and innovationHelp enable full automationAccess convergenceScalability up and downHelp enable end-to-end orchestrationMassive IoTImproved edge supportService flex
38、ibilityEnd-to-end slicingFully dynamic slicingEnhance DSSURLLC BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON83 Cloud-native networks are driving a shift from vertically integrated stacks to disaggregated cloud platforms 3.1 Operators must decide how to source cloud-native infr
39、astructure for 5GOperators that are planning their 5G SA core deployments and are looking to implement virtualized/Open RANs in the future must decide on their strategies for cloud-native infrastructure.Cloud-native clouds are horizontal in nature.They enable operators to shift away from verti-cally
40、 integrated stacks in which network functions are tightly bound to a vendors virtualized execution environment and towards open and disaggregated network cloud deployment models in which cloud-native infrastructure is sourced independently from vendors CNFs.Operators are showing a growing interest i
41、n building horizontal cloud-native network clouds,and we anticipate that disaggregated models will account for the largest share of spending on mobile network cloud software infra-structure in 2024.2Operators no longer face a straight-forward choice between sourcing cloud-native infrastructure as pa
42、rt of a vendor-integrated stack(the strategy that most operators followed in the NFV era)and building private horizontal network clouds from multiple disaggre-gated technologies.PCPs are introducing a further option:the ability to use a version of their cloud-native platforms that is tailored for te
43、lecoms workloads,either on-or off-premises or via a hybrid private/public deployment model.Each deployment model(which we call single-vendor-integrated,DIY and public cloud platform)has benefits and drawbacks,which we will briefly assess in the following sections.3.2 Vendor-integrated stacks limit f
44、lexibility and choice,but retain appeal for certain operatorsSingle-vendor,vertically integrated stacks have remained the predominant model for deploying virtualized network functions(VNFs)over the past 10 years,but they have not delivered the promised benefits of NFV.Most operators have experienced
45、 minor capex savings from moving to COTS hardware and little,if anything,in the way of opex savings because they continue to operate VNFs as virtualized appliances.However,this model continues to appeal to operators because it does not disrupt their opera-tional processes,deployment is quick due to
46、minimal integration challenges and operators only need deal with a single vendor for maintenance and support.On the other hand,the vertically integrated model has disadvantages for operators that want to control their own destinies.It results in vendor lock-in,2 For more information,see Analysys Mas
47、ons Network cloud infrastructure:worldwide forecast 20222027.BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON9which can limit operators access to the innovations and best-in-class functions that are being developed by a broader ecosystem.Operators therefore have fewer opportuniti
48、es to differentiate their network capabilities and services.Operators also fail to benefit from cloud economics because every function runs in its own siloed environment,hence the lack of opex savings.3.3 DIY private clouds require deep pockets to build and maintainOperators that want the benefits o
49、f an open,horizontal cloud that can support network functions from multiple vendors have traditionally built such clouds themselves from disaggregated compo-nents,including compute and storage hardware,cloud technology platforms,SDN and,in advanced cases,orches-tration capabilities.Operators have bu
50、ilt DIY private clouds to support their mobile packet cores and IMS,though few have demonstrated that they are disciplined enough to run these and other functions on the same network cloud.More commonly,operators have accrued multiple DIY private clouds,each dedicated to a single network function.Ev
51、en when operators have managed to create a multi-vendor DIY private cloud,they have dispersed its operations among multiple network function owners,thereby perpetuating functional silos and reducing the opportunities for opex savings.At the same time,operators have had to contend with the complex-it
52、ies of a disaggregated cloud environment in which different cloud technologies evolve at different rates and provide different approaches to automation.Building and maintaining integration and automation across all network cloud components(including network functions)imposes a continual overhead on
53、the operators that take this route.However,operators that master DIY private clouds believe that they have a strong opportunity to differentiate their networks and avoid relying on any single component supplier.3.4 PCPs offer a new approach to network clouds based on their hyperscale cloud technolog
54、ies and high levels of automationPCPs are now entering the network cloud market and are bringing their massive investments in cloud technol-ogies,automation frameworks and skills with them.They and their operator customers argue that cloud infra-structure is a commodity that does not provide competi
55、tive differentiation to a network owner,but that can be used to gain large economies of scale.PCPs offer managed services and very high levels of automation that minimize the challenges of building and operating disaggregated clouds.They also offer flexible deployment options:their cloud stacks can
56、run on-premises or via a hybrid private/public cloud model.BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON10However,PCP cloud deployments are still nascent.Operators also face potential data sovereignty and regulatory issues if PCPs do not operate their public clouds from multip
57、le in-country locations,and operators need a strategy to overcome the gap between the three nines of availability that PCPs typically provide and the carrier-grade five nines of availability expected by operators.3.5 Operators must understand the full cost implications of disaggregated deployment mo
58、dels to guide their cloud platform choicesCost/ROI is the top barrier to the adoption of 5G SA according to our 2022 operator survey(Figure 3.1).Operators regard cloud deployments as expensive given the uncertain ROI,especially those operators that have built disaggregated private clouds themselves
59、to support previous generations of VNFs.Such operators report little,if any,reduction in opex from these DIY cloud platforms because building and operating a private network cloud on-premises is highly complex and therefore consumes signif-icant time and resources.Operators must maintain large opera
60、tional headcounts to carry out slow,manual processes or build new automation solutions for their private clouds.Typically,operators do not understand the true extent of these costs.Analysys Mason has developed a robust TCO model to analyze the capex and opex of the various disaggregated network clou
61、d models and to reveal the complete set of costs associated with them in order to advise operators that are making disaggregated cloud platform decisions.FIGURE 3.1:BARRIERS TO 5G SA NETWORK ADOPTION,WORLDWIDE,20223 SOURCE:ANALYSYS MASON,20233 What are the main barriers for implementing a 5G SA core
62、?;n=68.0%5%10%15%20%25%30%Percentage of operatorsCost/ROIImmature technologyRobustness concernsLack of compelling use casesMulti-vendor solutions are immatureNSA meets our needsNo current need for advanced 5G capabilitiesDisruption of migration BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A T
63、CO COMPARISON11Analysys Mason,in collaboration with Microsoft,has analyzed the TCO of deploying a cloud-native 5G SA network using a PCP cloud solution based on Azure Operator Nexus compared with that using the DIY private cloud model.We created a set of models to assess the capex and opex associate
64、d with cloud-native mobile core networks and their operations.We then used these models and alongside inputs from Tier-1 operators that adopted the DIY private cloud model for their 5G SA networks to analyze the benefits of implementing a common,managed cloud platform and automation framework,such a
65、s Azure Operator Nexus,instead of using the DIY private cloud model.Our TCO model analyzes three greenfield deployment scenarios for the cloud-native 5G SA core in consumer macro networks,as detailed in Figure 4.1.These scenarios are modeled for two FIGURE 4.1:TCO MODEL DEPLOYMENT SCENARIOS SOURCE:A
66、NALYSYS MASON,20234 Azure Operator Nexus can reduce the TCO of 5G SA network cloud infrastructure by up to 43%compared to a DIY private cloudScenarioDeployment modelCloud software infrastructure(CaaS,PaaS,OS and SDN control)Cloud hardware infrastructure(compute,storage and network)Cloud infrastructu
67、re operations(Day 0,1 and 2)DIY private cloudAll functions are deployed on-premisesOperators build their own by procuring and integrating components from multiple vendorsOperators build their own by procuring and integrating components from multiple sourcesOperators are fully responsible for the man
68、agement of the entire lifecycleAzure Operator Nexus on-premisesAll functions are deployed on-premisesOperators implement Azure Operator Nexus(an integrated suite of cloud software infrastructure elements that is delivered as-a-service via a subscription model)Operator procures the hardware infrastru
69、cture directly from OEMs based on prescriptive BoM provided by AzureAzure provides managed services for the on-premises cloud infrastructure operationsAzure Operator Nexis hybridUser plane functions are deployed on-premises and some control plane and management functions are deployed on the public c
70、loudThe same Azure Operator Nexus platform is deployed both on-premises and in the Azure cloudOn-premises:same as abovePublic cloud:Azure cloudAzure provides managed services for hybrid cloud infrastructure operations BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON12hypothetical
71、 Tier-1 operator profiles:one from Western Europe and the other from Latin America.Each profile reflects the real-life subscriber numbers,network traffic and design and cost parameters for operators in these geographies.Key modeling assumptionsThe key modeling assumptions for the two operator profil
72、es are provided in Figure 4.2.5G SA core architecture and components(Figure 4.3)are modeled in FIGURE 4.2:KEY MODELING ASSUMPTIONS FOR THE OPERATOR PROFILES SOURCE:ANALYSYS MASON,2023FIGURE 4.3:KMODELING ASSUMPTIONS RELATED TO TECHNOLOGY AND ARCHITECTURE SOURCE:ANALYSYS MASON,2023AttributeTier-1 ope
73、rator in Western EuropeTier-1 operator in Latin AmericaCountry mixGermany,Spain and the UKBrazil and MexicoTotal number of 5G connections14 million in year 1 and 50 million in year 50.5 million in year 1 and 40.6 million in year 5Total 5G traffic4800PB in year 1 and 29 000PB in year 5375PB in year 1
74、 and 6750PB in year 5Total number of server nodes300(year 5)220(year 5)AttributeAssumptionModeled network componentsCloud-native 5G SA functions including AMF,SMF,UPF,NRF,NSSF,SDM and policy managementCapex parametersHardware,software and professional services for CNFs,network orchestration and elem
75、ent management systemsOpex parametersCloud software infrastructure(CaaS and OS)are assumed to be subscription-based for both the DIY private cloud and Azure Operator Nexus-based modelsCosts of public cloud infrastructure usage in the hybrid Azure Operator Nexus scenarioSupport and maintenance for so
76、ftware and hardware componentsHeadcount(FTE)hours for Day 0,1 and 2+operationsPower and spaceLength of analysis5 years BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON13the same way for both profiles,and labor costs reflect the regional variances for both the DIY private cloud an
77、d Azure Operator Nexus-based deployment scenarios.4.1 The hybrid Azure Operator Nexus model offers the lowest overall TCO of all scenariosFigure 4.4The cloud infrastructure and operations TCO of the shows the cumulative,5-year TCO of a cloud-native 5G SA network deployment for each of the three scen
78、arios.The overall TCO of the Azure Operator Nexus on-premises scenario is 24%lower than that of the DIY private cloud deployment for the Western European operator and 25%lower for the Latin American operator.Opex reduction enabled by the increased automation and efficiency of the cloud infrastructur
79、e and CNF operations is the main driver of the TCO savings in the Azure Operator Nexus scenario.Azure Operator Nexuss comprehensive and consistent cloud environment,automation framework and managed service capabilities can reduce the FTE working hour requirements for the main cloud-native network op
80、erations by 59%compared to the DIY private cloud model and can reduce opex by up to 36%.The operational efficiencies that can be achieved with Azure Operator Nexus are discussed in section 4.2.The hybrid cloud implementation of Azure Operator Nexus provides the optimal TCO of all scenarios because i
81、t enables further capex and opex savings compared to the Azure Operator Nexus on-premises scenario by allocating suitable workloads to the Azure public cloud.This helps operators to achieve higher scalability and utilization levels and reduce on-premises hardware costs,as discussed in section 4.3.FI
82、GURE 4.4:CUMULATIVE,5-YEAR TCO FOR THE THREE CLOUD-NATIVE 5G SA NETWORK DEPLOYMENT SCENARIOS SOURCE:ANALYSYS MASON,20230204060103050Azure Operator Nexus hybridAzure Operator Nexus on-premisesDIY private cloudAzure Operator Nexus hybridAzure Operator Nexus on-premisesDIY private cloudCapexOpexLatin A
83、mericaWestern EuropeTCO(USD million)BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON144.2 The cloud infrastructure and operations TCO of the Azure Operator Nexus on-premises scenario is up to 38%lower than that for the DIY model due to a reduction in opexFigure 4.5 and Figure 4.6
84、 show the cloud infrastructure and operations TCO(which excludes CNF-related costs)for the two on-premises deployment scenarios for a Tier-1 operator in Western Europe and in FIGURE 4.5:CLOUD INFRASTRUCTURE AND OPERATIONS TCO FOR THE ON-PREMISES MODELS,WESTERN EUROPE SOURCE:ANALYSYS MASON,2023FIGURE
85、 4.6:CLOUD INFRASTRUCTURE AND OPERATIONS TCO FOR THE ON-PREMISES MODELS,LATIN AMERICA SOURCE:ANALYSYS MASON,2023TCO componentDIY private cloudAzure Operator Nexus on-premisesComparison of Azure with DIYCapexHardware(compute,storage and network)including implementation and integration servicesUSD11 6
86、50 000USD10 380 00011%Total capexUSD11 650 000USD10 380 00011%OpexCloud platformUSD3 536 842USD4 502 920+27%Hardware support and maintenanceUSD5 384 784USD4 779 66811%Labor(FTE hours)USD30 850 000USD12 820 00058%Power and spaceUSD1 680 000USD1 680 0000%Total opexUSD41 451 627USD23 782 58843%Total TC
87、O(5-year)USD53 101 627USD34 162 58836%TCO componentDIY private cloudAzure Operator Nexus on-premisesComparison of Azure with DIYCapexHardware(compute,storage and network)including implementation and integration servicesUSD6 530 000USD5 440 00017%Total capexUSD6 530 000USD5 440 00017%OpexCloud platfo
88、rmUSD1 656 782USD2 094 582+26%Hardware support and maintenanceUSD3 003 615USD2 479 23917%Labor(FTE hours)USD15 430 000USD6 400 00059%Power and spaceUSD420 000USD420 0000%Total opexUSD20 510 397USD11 393 82144%Total TCO(5-year)USD27 040 397USD16 833 82138%BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NE
89、TWORKS:A TCO COMPARISON15Latin America,respectively.The cloud infrastructure and operations TCO for the Azure Operator Nexus scenario is 36%and 38%lower than that for the DIY private cloud in Western Europe and Latin America,respectively.The slight difference in TCO savings between the operator prof
90、iles is mainly caused by the greater scalability and cost advantages of the Azure Operator Nexus architecture and subscription-based model in Latin America because this deployment starts from a very small base and scales up quickly over 5 years.Our TCO model shows that the capex of the on-premises A
91、zure Operator Nexus scenario is1117%lower than that of the DIY private cloud model.Both scenarios are based on industry-standard COTS hardware components with similar specifications and costs,but Azure Operator Nexus enables indirect capex savings thanks to its predesigned and validated infrastructu
92、re.This leads to a major reduction in the cost of cloud hardware implementation,integration and testing.One of the key findings from our TCO analysis is that there is significant amount of opex attached to the lifecycle operations of cloud infrastructure and CNFs,and operators are not usually fully
93、aware of the size of these costs.Our research and interviews show that operators are struggling with the private cloud and on-premises data center management and automation,and are burdened by the cost of large operational headcounts.This is due to:unfamiliarity with cloud-native technologies and a
94、lack of in-house expertise to operate and automate cloud-native networks highly fragmented and not-fit-for purpose management and automation tools and slow,manual data center processes different automations for different network functions and for between infrastructure and network functions,which re
95、sults in limited economies of scale,integration issues and the need for complex network orchestration mechanisms.We collected granular,real-life data for headcounts and FTE hours for each of the major cloud-native network opera-tional processes from several large Tier-1 operators to examine the true
96、 costs of operating DIY private clouds and to enable a complete TCO assessment.Our data and analysis show that FTE hours are the biggest contributor to the total cost of a 5G SA network deployed using a DIY private cloud and can account for up to 40%of the overall TCO.One of the main benefits of the
97、 Azure Operator Nexus cloud platform is that it provides operators with a comprehensive set of integrated cloud automation tools,PaaS and managed services expertise in order to enable users to achieve highly automated cloud-native network opera-tions with significantly fewer FTE hours.Conversely,ope
98、rators that take the DIY approach need to stitch together all of these components by themselves,which BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON16is highly complex and involves significant risks.They must also perform their own lifecycle management,which is opex-heavy.For e
99、xample,AT&T built its Airship cloud using the DIY model,but was unable to get it to function like a public cloud.Other CSPs have employed multiple vendors/systems integrators to build their cloud platforms and automa-tions,but each time they have ended up reinventing the wheel for themselves and hav
100、e built unique cloud automations that cannot be reused for other functions.Our model shows that Azure Operator Nexus scenario can reduce the opex associated with FTE hours by up to 59%.It can also reduce the opex associated with cloud infrastructure and CNF opera-tions by 43%in Western Europe and 44
101、%in Latin America.Most of the opex savings enabled by the Azure Operator Nexus model come from the ability to automate and remotely manage cloud infrastructure operations,as shown in Figure 4.7.Figure 4.8 illustrates the cumulative breakdown of the FTE hour cost reductions.In addition to cloud infra
102、structure,Azure Operator Nexus can also help operators with the automated management of several key CNF operational processes,especially when the operator streamlines and simplifies its CNF platform services environment using Azure-managed PaaS solutions(such as logging,monitoring,AI/ML,CI/CD and De
103、vOps tools).Operators can achieve the FTE hour reductions in CNF operations that are shown in Figure 4.9 by following this approach.Conversely,operators that use the DIY model implement fragmented,vendor-specific platform services and tools and manage their lifecycle by themselves.Figure 4.10 provid
104、es a cumulative breakdown of Azure Operator Nexus FTE hour cost reductions by each CNF process.4.3 Adopting a hybrid cloud architecture based on Azure Operator Nexus could result in a cloud infrastructure and operations TCO reduction of 43%We assessed the capex and opex impli-cations of using hybrid
105、 cloud architecture for a cloud-native 5G SA network(Azure Operator Nexus hybrid),where some of the control plane functions(AMF and policy management)and management applications(network orchestration and element managers)are deployed in the Azure public cloud and the rest of the network functions ar
106、e deployed on-premises in operators data centers.Our model shows that this scenario can reduce the TCO compared to DIY model as follows.The overall TCO for a Western European operator is 26%lower(compared to 24%for the on-premises model)and the cloud infrastructure and operations TCO is 38%lower(com
107、pared to 36%for the on-premises model).The overall TCO for a Latin American operator is 28%lower(compared to 25%for the on-premises model)and the cloud infrastructure and operations TCO is 43%lower(compared to 38%for the on-premises model).BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO CO
108、MPARISON17FIGURE 4.7:OPEX BENEFITS RELATED TO CLOUD INFRASTRUCTURE OPERATIONAL PROCESSES OF THE AZURE OPERATOR NEXUS MODEL COMPARED TO THE DIY MODEL SOURCE:ANALYSYS MASON,2023Operational processBenefits of the Azure Operator Nexus model over the DIY modelPotential reduction in FTE hoursMean time to
109、repair(MTTR)Azure Operator Nexus provides closed-loop automation with real-time analytics/AI for the monitoring,root-cause identification and remediation of the entire cloud infrastructure.This is fully managed by operators in the DIY private cloud model and requires the equivalent of 7 FTEs annual
110、working hours per year.Fewer than 1.5 FTEs annual working hours are required per year for the Azure Operator Nexus model.80%SecOpsAzure Operator Nexus provides full security lifecycle management and expertise including security monitoring,incident response,proactive identification and fixing of vuln
111、erabilities,roll-outs of security updates/patches and tests.This requires only around 1.6 FTEs annual working hours per year.70%Network deployment and provisioningAzure Operator Nexus provides a Day 0 and 1 zero-touch automated deployment model for cloud infrastructure resources and ensures that the
112、 correct OS features and firmware are installed and provisioned.In a DIY private cloud environment,there are usually many moving parts that come from multiple vendors and a significant amount of engineering work is required to deploy and validate the infrastructure.Azure Operator Nexus provides simp
113、lified and consistent hardware infrastructure BoM based on a pre-certified design of compute,storage and networking resources,which reduces the design costs and risks of inconsistent DIY environments.It also accelerates hardware procurement.Logistics and systems integrator partners of Azure Operator
114、 Nexus put everything together and deliver it to the operator,thereby reducing rack/stack/cabling efforts.75%SysAdminOn-going lifecycle management processes such as server management,OS/firmware upgrades and patching are delivered remotely by Microsoft,thereby reducing the need for dedicated SysAdmi
115、n FTEs.80%Cloud configurationAzure Operator Nexus provides automated configuration and validation of cloud software(Kubernetes/CaaS,OS)and hardware resources.Azure Operator Nexus includes advanced fabric automation with fully integrated SDN capabilities to connect physical and logical networks to th
116、e cloud.This is a highly complex process and DIY operators usually do this in a manual way using spreadsheets and scripts.70%CI/CD pipelineThe coherent cloud infrastructure components of the Azure Operator Nexus platform enables streamlined and efficient CI/CD processes that are supported by Azure O
117、perator Nexus PaaS and CI/CD tools compared to siloed DIY cloud infrastructure environments.80%Monitoring and observability set-upCloud hardware,software and applications are all in the same Azure Operator Nexus environment and come with curated,out-of-the box monitoring and observability capabiliti
118、es.DIY clouds usually consist of different set-ups that require manual stitching.This can be error-prone,slow and costly to carry out.80%BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON18FIGURE 4.8:BREAKDOWN OF AZURE OPERATOR NEXUS FTE HOUR COST SAVINGS BY CLOUD INFRASTRUCTURE OP
119、ERATIONAL PROCESSES SOURCE:ANALYSYS MASON,2023FIGURE 4.9:OPEX BENEFITS RELATED TO CNF OPERATIONAL PROCESSES OF THE AZURE OPERATOR NEXUS MODEL COMPARED TO THE DIY MODEL SOURCE:ANALYSYS MASON,20230%5%10%15%20%25%30%35%40%Share of FTE hour cost savingsMTTRSecOpsTesting and integrationNetwork deployment
120、 and provisioningSysAdminMonitoring and observability set-upCloud configurationCI/CD pipelineOperational processBenefits of the Azure Operator Nexus model over the DIY modelPotential reduction in FTE hoursxNF onboarding and deploymentAzure Operator Nexus hardware and software pretesting and certific
121、ation can reduce the effort and costs of CNF onboarding processes.Azure Operator Nexus also works with a large ecosystem of CNF vendor partners whose CNF images and Helm charts are onboarded to a standard marketplace catalog and plugged into Azure PaaS.40%SecOpsThe Microsoft security team performs t
122、he full security lifecycle of CNFs on behalf of the operator and the operator therefore only needs to allocate limited working hours to oversee the processes.70%CI/CD pipelineAzure Operator Nexus technology(PaaS and GitHub)and expertise is delivered in an integrated and efficient manner,whereas DIY
123、environments are usually a patchwork of multiple pipelines and technology islands,which increases the cost and complexity of CI/CD pipeline set-ups.50%Monitoring and observability set-upAzure Operator Nexus provides a unified observability platform that includes a coherent set of out-of-the-box comp
124、onents and a single pane of glass for CNFs and cloud infrastructure.DIY approaches usually lead to fragmented and complex monitoring and observability environments that are more expensive to build and manage.The savings here are lower than those related to cloud infrastructure because many CNFs stil
125、l depend on their proprietary,closed element management systems(EMS),which restricts the automation capabilities.50%BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON19Figure 4.11 shows the capex and opex savings of this hybrid architecture for a Western European operator(see Figur
126、e 6.6 in the annex for the data for a Latin American operator).These savings mainly come from the following factors and can potentially be enhanced by further optimizing the workload allocation by using the public cloud environment more.The reduced on-premises hardware footprint and associated costs
127、 lead to lower hardware capex and related opex(such as energy and support and maintenance costs).A portion of the hardware infrastructure investment shifts from high up-front capex to consumption-based opex,which delivers an optimized cost model.The use of public cloud infrastructure eliminates the
128、need for overprovisioning hardware for peak times,which leads to higher utilization and scalabilityFIGURE 4.10:BREAKDOWN OF AZURE OPERATOR NEXUS FTE HOUR COST SAVINGS BY CNF OPERATIONAL PROCESSES SOURCE:ANALYSYS MASON,20230%10%20%30%40%50%60%70%Share of FTE hour cost savingsSecOpsTesting and integra
129、tionxNF onboarding and deploymentCI/CD pipelineMonitoring and observability set-up BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON20FIGURE 4.11:CLOUD INFRASTRUCTURE AND OPERATIONS TCO FOR ALL THREE DEPLOYMENT SCENARIOS,WESTERN EUROPE SOURCE:ANALYSYS MASON,2023TCO componentDIY pr
130、ivate cloudAzure Operator Nexus on-premisesAzure Operator Nexus hybridComparison of Azure hybrid with DIYCapexHardware(compute,storage and network),including implementation and integration servicesUSD11 650 000USD10 380 000USD6 520 00044%Total capexUSD11 650 000USD10 380 000USD6 520 00044%OpexCloud
131、platformUSD3 536 842USD4 502 920USD3 039 04514%Hardware support and maintenanceUSD5 384 784USD4 779 668USD2 835 53147%Labor(FTE hours)USD30 850 000USD12 820 000USD12 820 00058%Public cloud usage00USD6 608 393N/APower and spaceUSD1 680 000USD1 680 000USD1 170 00030%Total opexUSD41 451 627USD23 782 58
132、8USD26 472 96936%Total TCO(5-year)USD53 101 627USD34 162 588USD32 992 96938%BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON215 Conclusions and recommendations Cloud-native infrastructure is critical to enable operators to achieve the benefits that they expect to derive from depl
133、oying a 5G SA core,including the ability to support advanced network services.Cloud-native infrastructure enables operators to shift away from vertically integrated network function stacks to more open and disaggregated network cloud deployment models that support access to innovation within a large
134、r network function vendor ecosystem.Operators need to understand the complete set of costs associated with the various approaches to sourcing a disag-gregated cloud platform.This will allow them to make an informed decision about which cloud deployment model they should use to futureproof their 5G i
135、nfrastructure(both core and RAN).Our TCO study shows that there is substantial amount of opex attached to DIY private clouds due to the cloud-native automation challenges outlined in this paper.This can be detrimental to operators ability to deliver automated and competitive 5G and edge services.A c
136、omprehensive and coherent cloud-native platform and automation framework delivered via an as-a-service model(such as Azure Operator Nexus)can help operators to significantly reduce their private cloud opex and optimize the TCO of a cloud-native 5G SA network implementation.In addition,hybrid cloud a
137、rchitecture can potentially offer a more flexible and scalable network environment,which can improve the TCO/business case for cloud-native networks.We provide the following recommenda-tions for operators that are evaluating cloud-native 5G SA network deployment models and architecture to identify t
138、he optimal implementation approaches.Operators should choose the right cloud platform and operational model for their 5G SA networks in order to achieve their service and automation ambitions.Cloud infrastructure is increasingly critical for the network and for 5G monetization opportunities,so it ne
139、eds to be state-of-the-art,operationally efficient and delivered and managed using industry best practices.Operators that want to build true cloud environments for their 5G networks should embrace open,disaggregated network cloud models.These networks should be based on proven cloud-native platforms
140、 and automation frameworks that minimize the costs and challenges of building and operating disaggregated clouds and deliver programmable infrastructure to maximize service agility and innovation within a strong,open ecosystem.Operators should assess their operational readiness and in-house capabili
141、ties in order to conduct a complete TCO analysis.It is difficult and BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON22time-and resource-consuming for many operators to build and operate cloud-native networks on private clouds.Operators need to undertake a detailed examination of
142、 their cloud and data center processes and in-house operational capabilities to reveal their hidden costs and operational challenges.This will enable operators to build a more-holistic TCO/business case based on automation that will guide their technology vendor and cloud delivery model investment d
143、ecisions.Operators should devise a long-term plan to streamline and converge their existing and new clouds onto common cloud platforms and automation frameworks.Several advanced operators have taken the DIY path to build their disaggregated network clouds,but their progress has been stalled due to t
144、echnical and operational challenges(such as integration and automation/orchestration complexity and a lack of in-house skills and expertise).They have accumulated multiple network cloud silos using various tightly integrated xNFs and clouds,and have built specific automations that are not transferra
145、ble to other clouds.Operators that are starting to implement 5G SA and vRAN/Open RAN need to have a long-term plan to streamline and converge these clouds using common cloud platforms and industry-standard PaaS and cloud-native automation frameworks.Our TCO analysis shows that using the Azure Operat
146、or Nexus platform and its as-a-service model is a highly cost-efficient way of building new cloud-native networks,and operators could consider extending it to other cloud domains,including their existing network cloud silos.Operators should consider adopting hybrid cloud architecture for their cloud
147、-native networks.Our TCO analysis demonstrates that hybrid cloud architecture could deliver additional cost savings with commercial and deployment flexibility benefits.However,this option is still relatively new to network clouds,and operators should first follow a step-wise approach starting from l
148、ow-hanging fruit functions and use cases(such as those related to the control plane and policies).Operators that take this path must carefully evaluate their vendor and cloud platform choices to ensure that they meet their performance/QoS,data sovereignty/privacy,security and support/SLA requirement
149、s.BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON236 Annex This section provides details of the modeling assumptions and parameters of the TCO analysis.6.1 Connections and traffic6.2 Cost parameters The costs for training and OSS/BSS integration for 5G SA core network components
150、 are excluded in all scenarios.CNF costs are normalized and averaged from various vendor price benchmarks used in Analysys Masons consulting projects and regulatory models.Cloud infrastructure software(CaaS,SDN and OS)and COTS hardware costs are collected and normalized from various internal sources
151、 including Analysys Masons consulting projects,regulatory models and operator surveys.Azure Operator Nexus and Azure public cloud usage costs are estimated based on Microsofts guidance.FTE-related opex data was collected from Analysys Masons survey of five Tier-1 operators that implemented a 5G SA c
152、ore using the DIY model.The data includes the real-life frequency and duration of each operational activity and the number of FTEs involved.It has been normalized and averaged across these five operators.FIGURE 6.1:TOTAL NUMBER OF 5G CONNECTIONS FOR EACH OPERATOR PROFILE SOURCE:ANALYSYS MASON,2023FI
153、GURE 6.2:ANNUAL 5G DATA TRAFFIC(PB)FOR EACH OPERATOR PROFILE SOURCE:ANALYSYS MASON,2023Year 1Year 2Year 3Year 4Year 5Tier-1 Western European operator13 853 176 23 177 316 33 170 399 42 615 026 50 119 043 Tier-1 Latin American operator466 243 2 657 645 10 954 262 24 541 296 40 633 546 Year 1Year 2Yea
154、r 3Year 4Year 5Tier-1 Western European operator4765 9533 16 048 22 915 29 266 Tier-1 Latin American operator375 889 2044 4141 6752 BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON24 The lifetime of CNF licenses is assumed to be 5 years and COTS servers lifetimes are assumed to be
155、 3 years.Cloud hardware installation and provisioning costs are assumed to be 25%of the total hardware costs.Support and maintenance costs for all hardware elements are assumed to be 20%of the equipment costs.Support and maintenance costs for all software elements are assumed to be 14%of the softwar
156、e license costs.FIGURE 6.3:SUMMARY OF LABOR COST PARAMETERS SOURCE:ANALYSYS MASON,2023FIGURE 6.4:SUMMARY OF POWER COST PARAMETERS SOURCE:ANALYSYS MASON,2023FIGURE 6.5:SUMMARY OF FLOORSPACE COST PARAMETERS SOURCE:ANALYSYS MASON,2023ParameterValueRack units per standard rack42Rack floorspace(m2/rack)4
157、Cost trend of floorspace3%Annual floorspace cost per square meter Western EuropeUSD3000Annual floorspace cost per rackspace Western EuropeUSD12 000Annual floorspace cost per square meter Latin AmericaUSD1500Annual floorspace cost per rackspace Latin AmericaUSD6000ParameterValueHourly cost of labor W
158、estern EuropeUSD100 Hourly cost of labor Latin AmericaUSD50Cost trend of labor2.0%Number of working hours per year1725 ParameterValuePower(at a consumption of 48V per kWh)Western EuropeUSD0.10Power(at a consumption of 48V per kWh)Latin AmericaUSD0.05 BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWOR
159、KS:A TCO COMPARISON256.3 TCO breakdown for Tier-1 operator in Latin America for all three deployment scenariosFIGURE 6.6:CLOUD INFRASTRUCTURE AND OPERATIONS TCO FOR ALL THREE DEPLOYMENT SCENARIOS,LATIN AMERICA SOURCE:ANALYSYS MASON,2023TCO componentDIY private cloudAzure Operator Nexus on-premisesAz
160、ure Operator Nexus hybridComparison of Azure hybrid with DIYCapexHardware(compute,storage and network),including implementation and integration servicesUSD6 530 000USD5 440 000USD3 050 00053%Total capexUSD6 530 000USD5 440 000USD3 050 00053%OpexCloud platformUSD1 656 782USD2 094 582USD1 371 71717%Ha
161、rdware support and maintenanceUSD3 003 615USD2 479 239USD1 118 73963%Labor(FTE hours)USD15 430 000USD6 400 000USD6 400 00059%Public cloud usage00USD3 269 707N/APower and spaceUSD420 000USD420 000USD270 00036%Total opexUSD20 510 397USD11 393 821USD12 430 16439%Total TCO(5-year)USD27 040 397USD16 833
162、821USD15 480 16443%BUILDING COST-EFFICIENT CLOUD-NATIVE 5G SA NETWORKS:A TCO COMPARISON26Gorkem Yigit(Principal Analyst)is the lead analyst for the Cloud Infrastructure Strategies and Multi-Cloud Networking research programmes.His research focuses on the building blocks,architecture and adoption of
163、the cloud-native,disaggregated and programmable digital infrastructure and networks that underpin the delivery of 5G,media and edge computing services.He also works with clients on a range of consulting projects such as market and competitive analysis,business case development and marketing support
164、through thought leadership collateral.He holds a cum laude MSc degree in economics and management of innovation and technology from Bocconi University(Milan,Italy).Caroline Chappell(Research Director)heads Analysys Masons Cloud research practice.Her research focuses on service provider adoption of c
165、loud to deliver business services,support digital transformation and re-architect fixed and mobile networks for the 5G era.She is a leading exponent of the edge computing market and its impact on service provider network deployments and new revenue opportunities.She monitors public cloud provider st
166、rategies for the telecoms industry and investigates how key cloud platform services can enhance service provider value.Caroline is a leading authority on the application of cloud-native technologies to the network and helps telecoms customers to devise strategies that exploit the powerful capabiliti
167、es of cloud while mitigating its disruptive effects.Luwen Meng(Consultant)is based in our Cambridge office.At Analysys Mason,she has worked on a variety of project types across a number of telecoms verticals,including supporting regulators,network operators,and vendors.Luwen has been active in model
168、ling,report writing and desk research,supporting a range of clients based both in Europe and internationally.Luwen has a PhD in Chemical Engineering from the University of Cambridge.7 About the author Published by Analysys Mason LimitedNorth West Wing,Bush HouseAldwychLondon WC2B 4PJUKTel:+44(0)20 7
169、395 Registered in England and Wales No.5177472 Analysys Mason Limited 2020We have used reasonable care and skill to prepare this publication and are not responsible for any errors or omissions,or for the results obtained from the use of this publication.The opinions expressed are those of the author
170、s only.All information is provided“as is”,with no guarantee of completeness or accuracy,and without warranty of any kind,express or implied,including,but not limited to warranties of performance,merchantability and fitness for a particular purpose.In no event will we be liable to you or any third pa
171、rty for any decision made or action taken in reliance on the information,including but not limited to investment decisions,or for any loss(including consequential,special or similar losses),even if advised of the possibility of such losses.We reserve the rights to all intellectual property in this p
172、ublication.This publication,or any part of it,may not be reproduced,redistributed or republished without our prior written consent,nor may any reference be made to Analysys Mason in a regulatory statement or prospectus on the basis of this publication without our prior written consent.Analysys Mason Limited and/or its group companies 2023.Stay connectedYou can stay connected by following Analysys Mason via Twitter,LinkedIn,YouTube or RSS feed.AnalysysMason