Preparing the packet, optical and fixed access networks ... · PDF fileAbout Huawei’s...

White paper

Preparing the packet, optical and

fixed access networks for the

digital era: requirements for a

cloud network platform

November 2017

Gorkem Yigit and Caroline Chappell

.

Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | i

© Analysys Mason Limited 2016 Contents

Contents

1. Executive summary 1

2. Starting the journey to a cloud-based network 3

2.1 Drivers for the cloud-based network 3

2.2 What is the cloud-based network? 5

2.3 Challenges and migration path 6

3. The cloud-based management of the packet, optical and fixed access networks 9

3.1 Drivers and benefits 9

3.2 An ideal multi-layer SDN platform will pave the way to DNOP 9

4. About Huawei’s NCE in cloud-based networks 12

4.1 Huawei’s primary goal is achieving network intelligence 12

4.2 NCE is a cloud-based, autonomous network management platform 12

4.3 NCE aims to provide a smooth evolution to autonomous networks 14

5. Conclusion 15

About the authors 17

About Analysys Mason 18

List of figures

Figure 1: WAN SDN automation as one of the key converging components of the DNOP framework

[Source: Analysys Mason, 2017] ...................................................................................................................... 1

Figure 2: Short-term and long-term business drivers for virtualised networks [Source: Analysys Mason,

2017] ................................................................................................................................................................. 4

Figure 3: Transformation of network operations from vertical management silos to horizontal platforms with

SCN [Source: Analysys Mason, 2017] ............................................................................................................. 6

Figure 4: Overview of the new operational stack to enable agile and automated operations of hybrid

networks [Source: Analysys Mason, 2017] , ..................................................................................................... 8

Figure 5: High-level functional blocks of multi-layer SDN control platform [Source: Analysys Mason, 2017]

........................................................................................................................................................................ 10

Figure 6: Functional capabilities of multi-layer SDN control and management platforms [Source: Analysys

Mason, 2017] .................................................................................................................................................. 10

Figure 7: The three essentials of building network intelligence [Source: Huawei, 2017] .............................. 12

Figure 8: Huawei NCE architectural overview [Source: Huawei, 2017] ........................................................ 13

Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 1

© Analysys Mason Limited 2017 Executive summary

1. Executive summary

This white paper outlines the key drivers and requirements for building cloud-based packet, optical and fixed

access (IP, MPLS, optical, microwave and fibre access) networks for communications service providers (CSPs).

The paper provides a definition of the cloud-based networks and operations based on Analysys Mason’s digital

network and operations platform (DNOP) vision;1 discusses the main challenges surrounding these networks

and operations; and recommends a migration path for CSPs that can minimise the complexity and risks of

software-defined networking (SDN) and network functions virtualisation (NFV)-driven transformations. It then

analyses the potential benefits of adopting a platform-based approach to implementing wide-area network

(WAN) SDN2 automation and identifies the desired characteristics and functionalities of an ideal multi-layer

SDN platform and how it fits into the wider DNOP framework, as illustrated in Figure 1 below.

Figure 1: WAN SDN automation as one of the key converging components of the DNOP framework [Source: Analysys

Mason, 2017]

1 For more details on the DNOP, see Defining the digital network and operations platform for 5G and future networks.

2 Analysys Mason’s WAN SDN definition encompasses SDN-controller driven and SDN-like deployments in CSPs’ IP/optical

access, metro and core WANs. For more details, see Software-defined networking (SDN) in the WAN: solution options and

vendor opportunities report.

Digital

network and

operations platform

(DNOP)

NFV

Cloud-based

orchestration and

network function

WAN SDN

Programmable

connectivity

management and

control

Operational

automation

End-to-end service

abstraction and

lifecycle

management

http://www.analysysmason.com/Research/Content/Short-reports/Defining-DNOP-5G-RMA16-RMA07-RMA18-RMA01-RMA02-RMA04/

http://www.analysysmason.com/Research/Content/Reports/SDN-WAN-strategy-Mar2017-RMA07



© Analysys Mason Limited 2017 Executive summary

Cloud-based networks and operations will be key to realise CSP transformation goals

Cloud-based networks will play a central role in supporting CSPs’ digital service provider (DSP) vision, which

is driven by new revenue opportunities, automation-led cost transformation and market differentiation. Building

future cloud-based networks will require a paradigm shift from a CSP’s traditional network with its discrete

physical appliances and multiple legacy operations support systems (OSSs) and element/network management

systems (E/NMSs) towards a coherent DNOP. The DNOP is the software architecture on which future networks,

such as 5G, will be based (see Section 2.2) Leading CSPs are taking the first steps towards building DNOPs

through industry collaborations, open-source development and multi-vendor initiatives. However, this will be a

step-by-step transformation journey over the next ten years or so for CSPs and their suppliers.

CSPs should follow a business-driven, gradual migration strategy to cloud-based networks with a

coherent WAN SDN platform at the core

CSPs cannot take a ‘big bang’ approach3 to cloud-based networks. This transformation needs to be applied in

domains based on CSPs’ business and operational needs. Most CSPs are choosing the enterprise services

domain, specifically L0-3 services such as leased lines and L2/L3 VPNs, because of competitive pressures.

However, these CSPs are using disparate tools and technologies when implementing WAN SDN (i.e. investing

in SD-WAN to grow revenue, modernising IP/MPLS/optical networks and joining these two together). In doing

so, CSPs run the risk of creating new silos that are challenging to integrate and that add cost. To avoid repeating

past mistakes, CSPs will need a common platform that can help them draw together separate connectivity

domains and ease integration and maintenance challenges (see Section 2.3).

Multi-layer SDN platforms should converge management and control to lead the way to cloud-based

network operations

Introducing SDN-based programmability across multiple network domains and layers will be a slow and

disruptive process. To fully realise the benefits of cloud-based packet/optical transport and fixed access

networks, CSPs will need to adopt open, modular and cloud-native multi-layer control platforms that can help to

disband existing single-vendor E/NMS silos by providing end-to-end, automated connectivity management and

control in multi-vendor WAN environments (see Section 3.2). This convergence of network management and

control in the multi-layer SDN platforms will enable CSPs to reshape their existing network operations stacks

(especially when combined with service and network orchestration), using new and more agile operational

platforms such as DNOP.

3 Where CSPs transform their entire business and their systems simultaneously.


© Analysys Mason Limited 2017 Starting the journey to a cloud-based network

2. Starting the journey to a cloud-based network

2.1 Drivers for the cloud-based network

Communication service providers (CSPs) are faced with the challenge of digital relevancy. Their traditional

communications services are increasingly being commoditised, putting pressure on CSPs’ revenue and margins, while

high-growth digital services such as cloud, video and the Internet of Things (IoT) are increasingly controlled by

FANG4 companies. CSPs’ current mode of operations and underlying systems cannot deliver the same level of

experience consumers are accustomed to receiving from these alternative service providers. Mass market consumers,

for example, are switching to highly personalised, innovative and cost-competitive communications and other digital

services offered by over-the-top (OTT) players such as Netflix, Facebook, WhatsApp and WeChat. Enterprises are

demanding real-time, on-demand connectivity services through self-care portals, similar to those delivered by

hyperscale cloud providers (e.g. AWS, Google Cloud, Microsoft Azure).

Leading CSPs are embracing software-controlled networking (SCN) technologies (SDN, NFV5 and cloud

computing) to maintain their relevance in this cloud-based, software-centric ICT world. These CSPs’ aspirations

for successful transformation to digital service providers (DSPs) include SCN technologies, together with

leaner, more automated operations that support new digital services and a digital customer experience.

Many CSPs’ virtualisation strategies, however, are tactical and defensive in the early stages, as illustrated in

Figure 2. These strategies focus on achieving cost reduction and improving service agility, for example by

adding a virtualised EPC to bring about hardware savings in capacity augmentations, and a virtualised IP

multimedia subsystem (vIMS) for faster introduction of VoLTE services. As SDN technologies mature, CSPs

are introducing SD-WAN and on-demand connectivity services for their enterprise customers, and using WAN

configuration tools to accelerate service provisioning, following AT&T’s success with Network on Demand.

Longer term, if CSPs wish to transform themselves into DSPs, they will need to look beyond these initial drivers

and use SCN to achieve the following business and operational benefits:

• Revenue growth with new SCN-enabled services, for example through on-demand enterprise and cloud

connectivity services bundled with virtual network functions (VNFs) delivered as a managed service.

Analysys Mason’s virtual customer premises equipment (vCPE) business case report6 demonstrates that a

developed market CSP can generate USD1.4 billion in new revenue from upselling and cross-selling

existing and new NFV/SDN-based services using vCPE and more programmable networks. In the next 3–5

years, the launch of 5G networks means that these new service areas will include smart cities, industry 4.0,

ultra-low latency remote device control, AR/VR video and self-driving cars.

• Cost transformation, beyond immediate capex/opex savings on a per use-case basis. CSPs should aim to

achieve extreme process automation and lean operations, which can be realised after a complete

virtualisation-driven operational and organisational transformation. Such a transformation will draw heavily

on artificial intelligence (AI) and machine-learning technologies, replacing human operators with

autonomic operations.

4 Facebook, Amazon, Netflix and Google: the term ‘FANG’ is increasingly used to describe any platform-based DSP.

5 SDN = software-defined networking; NFV = network functions virtualisation.

6 http://www.analysysmason.com/Research/Content/Reports/NFV-SDN-business-cases-Feb2016-RMA16/.



• Differentiation based on software mastery – as the network becomes more and more ‘software-ised’7 CSPs

will need to foster strong in-house software development capabilities if they are to control its evolution. They will

also need to competitively differentiate their network services from those of their competitors and attract an

ecosystem of application partners to create value using their network platforms. CSPs will need the same DevOps

culture and methodologies and open-source expertise as demonstrated by the FANG companies (Facebook,

Amazon, Netflix and Google), in order to uniquely position their network capabilities.

Figure 2: Short-term and long-term business drivers for virtualised networks [Source: Analysys Mason, 2017]

To realise these benefits, CSPs will need to disrupt current telecoms company mindsets, and transform closed

and inflexible networks and manual processes into new, more agile cloud-based networks and operations

enabled by SDN, NFV and cloud technologies.

7 ‘Network software-isation’ is an IEEE/5G term referring to the introduction of NFV/SDN.

Survive/defend revenue

Cost optimisation

Reduced capex

with standardised,

multi-tenant

infrastructure

Reduced opex

with increased

efficiency

Faster and cost-effective delivery of new,

on-demand, customer self-provisioned services

using software-controlled networking

(e.g. SD-WAN)

CSP

short-term

5G, network as a service (NaaS),

network slicing, IoT, video, cloud

DSP

long-term

Lifecycle upgrades

Upgrade legacy, out-of-date networks and

systems with cloud-based, future-proof

solutions

New digital services revenue

Cost transformation

Differentiation based on

software mastery

Differentiate networks and services faster than the

competition with strong in-house, DevOps software

development and open-source expertise

5G

Extreme process automation,

streamlined systems and leaner

operations underpinned by software-

controlled networking, analytics and AI



2.2 What is the cloud-based network?

The cloud-based network is enabled by advanced SDN/NFV technologies. These networks will be built from software

components that are natively engineered for cloud, which means that they are highly scalable, reusable, loosely-

coupled and open. The cloud-based network will be built using DevOps processes and tool chains and will include

embedded, end-to-end, (near) zero-touch, closed-loop automation across key operational processes, including:

• network management, planning and optimisation

• customer self-provisioning

• performance management and assurance.

The cloud-based network will require a paradigm shift from a CSP’s traditional network with its discrete

physical appliances and multiple legacy operations support systems (OSSs) and element/network management

systems (E/NMSs) towards a coherent DNOP that will span networking technologies and domains. The DNOP

encompasses both the network itself, and its operations, as software code: it will embrace VNFs, programmable

connectivity and embedded management and control functions within the same platform and integration

framework. DNOPs will introduce a cloud-native architecture: the building block of the DNOP will be

microservices, including a high proportion of those developed by open-source communities. Individual CSPs

will curate the set of components (microservices) within their DNOPs according to their particular business and

operational objectives.

CSPs are currently a long way from implementing a DNOP (see Figure 3 below). Most are in the first phase of the

journey towards the cloud-based network, deploying monolithic VNFs and experimenting with SDN-based network

management at the backbone and core networks. While this approach delivers some cost savings and agility benefits,

it also runs the risk of creating new network silos without improving the end-to-end network visibility, control and

automation needed to drive network cost transformation, differentiation and revenue growth.

As a next step, leading CSPs are progressing on two fronts. First, they are adopting cloud infrastructure that runs

multiple monolithic VNFs, which then collectively require management and orchestration. Second, they are

introducing automation to the physical wide area network (WAN) which connects physical and virtual network

functions across their own distributed cloud infrastructure and that of their customers. Analysys Mason calls this

automation WAN SDN.8 These CSPs understand that an integrated SDN and NFV approach is needed as a

further step towards achieving their broader automation goals. However, they still face significant challenges,

including the need to integrate the siloed systems that manage different types of physical network function so

that they can participate in end-to-end automation. They also need to create programmatic interfaces to legacy

physical functions.

8 Analysys Mason’s WAN SDN definition encompasses SDN-controller driven and SDN-like deployments in CSPs’ IP/optical

access, metro and core WANs. For more details, see Software-defined networking (SDN) in the WAN: solution options and

vendor opportunities report.





Figure 3: Transformation of network operations from vertical management silos to horizontal platforms with SCN

[Source: Analysys Mason, 2017]

In the third, cloud-native phase, CSPs will fully harness the benefits of extreme automation and cost transformation.

CSPs will evolve early WAN SDN capabilities onto a single architecture platform that spans multiple networking

domains. This will contribute to operational cost transformation by disbanding current NMS silos and enabling the

programmatic and cost-efficient, end-to-end delivery of services. As more and more network functions are virtualised

in a cloud-native manner, Analysys Mason expects the WAN SDN (connectivity) and NFV (functional) aspects of the

network to converge within the DNOP. We are already seeing leading CSPs take the first steps towards building

DNOPs through industry collaborations, open-source development and multi-vendor initiatives such as ONAP (the

merger of AT&T’s ECOMP and OPEN-O initiatives). For CSPs and their suppliers, this will not be an immediate

transformation, but rather a step-by-step transformation journey over the next decade.

2.3 Challenges and migration path

CSPs’ transformation to cloud-based network infrastructures would involve the decomposition and migration of

existing OSS and E/NMS silos to new operational models. This is not a process that can be performed and

completed at a single point in time. Such a ‘big-bang’ approach would risk causing significant disruption to

CSPs’ processes and services, creating a highly unstable business environment. To lower the complexity and

risks, CSPs will need to move towards cloud-based networks incrementally in logical, domain-based steps. This

will enable CSPs to build DNOPs in parallel to their existing OSSs and develop operational confidence in each

step before embracing full automation. Then they can phase out legacy OSSs and E/NMSs.

CSPs deliver a variety of services over multiple network domains including mobile and fixed access;

aggregation and transport IP; and optical networks spanning mobile backhaul (fibre, microwave), metro and

PNF VNF

E/NMS

OSS OSS

MANO

Wireless … Virtual

Vertical, service/domain/vendor-

specific management silos

Hyper-converged infrastructure

Digital network operations

platforms (DNOP)

Horizontal, streamlined, single

management platform

Virtualisation Orchestration Cloud-native

Transformation journey over timeCSP DSP

PNF

E/NMS

OSS

Fixed

Digital

services

Software-defined, extreme automation

Differentiation

New services revenue



backbone and core networks. An early challenge for CSPs will be to decide which of these services/network

domains should be prioritised. CSPs should consider targeting domain automation based on their business goals,

identifying main pain points and inefficiencies and then selecting and implementing the solutions that will

resolve them. For example:

• CSPs with a strategic focus on the highly competitive enterprise market need to reduce lengthy and manual

provisioning of connectivity and communications services, including transport, VPNs, firewalls and VoIP,

from weeks/months to minutes/hours. They also need to deliver differentiated service-level agreements

(SLAs).

• CSPs wishing to improve network utilisation, deliver new on-demand bandwidth services over today’s highly

static packet/optical transport networks, and tap into cloud and data centre interconnectivity demands need to

modernise the management and control of their IP/MPLS and optical networks. As a next step, they want to

integrate new technologies into their packet/optical networks to differentiate themselves in the market.

• CSPs with ambitions to deliver new services, such as IoT/M2M (currently) and/or 5G network use cases (in

the next 2–3 years) may want to build new greenfield networks based on the latest cloud networking

technologies to achieve the highest levels of efficiency and drive profitable revenue growth from the outset.

In order to achieve these business goals, CSPs are currently procuring a variety of new tools and products in a

piecemeal way. These include WAN configuration tools for the large-scale, automated configuration of multi-

vendor network devices (routers and switches), and traffic steering mechanisms (SD-WAN) at the network edge

to drive new revenue and multi-layer control platforms to enable the coherent management of the network

across packet and optical layers. In doing so, CSPs run the risk of re-creating the network management silos and

interoperability/integration challenges of the past. However, they can avoid these problems if they look for

commonalities between the automation and operational functions in separate connectivity domains and start to

bring them together into a common, end-to-end control and management platform as described in Figure 4

below. The adoption of such a WAN SDN platform is a key step in the creation of a DNOP.



Figure 4: Overview of the new operational stack to enable agile and automated operations of hybrid networks

[Source: Analysys Mason, 2017] 9,10

A platform-based approach to cloud-based network development requires openness as a critical factor. Most

vendors are focused on building proprietary tools and solutions primarily for the management of their own, and

their ecosystem partner’s, solutions. The cloud-based network will be built using open components which can be

procured from a wide variety of suppliers, giving CSPs choice and flexibility. CSPs should therefore ensure that

their procurement of cloud-based network and operations components takes into account their level of openness.

For example, CSPs should try to use:

• solution components that are based on open-source components (i.e. ODL, ONOS) and/or vendor-

developed proprietary code that can be plugged into open-source platform frameworks, such as ONAP

• vendor-agnostic solutions that support common industry standard models, protocols and open

interfaces/APIs for northbound and southbound integrations across OSS/orchestration layers, incumbent

vendor E/NMSs and multiple third-party SDN control solutions

• cloud-native, micro-services-based VNFs that can be deployed in any NFV infrastructure (NFVI), and

integrated and orchestrated by any management and orchestration (MANO) solution

• solutions that are supported by a large ecosystem of partners and a broad market (e.g. other CSPs, industry

collaborations).

9 http://www.analysysmason.com/Research/Content/Reports/SDN-WAN-strategy-Mar2017-RMA07.

10 http://www.analysysmason.com/Research/Content/Reports/vNGN-OSS-framework-Sep2015-RMA16-RMA07.

Service orchestration

Cross-domain network orchestrator (CD NO)

NFVO

VNFM

VIM/DC SDN Domain-specific SDN controllers

Mobile

Core Aggregation Edge

Transport Customer premises

Data centre

VNFs

NFVI

WAN SDN cross-domain controller

Hybrid network management and control


http://www.analysysmason.com/Research/Content/Reports/vNGN-OSS-framework-Sep2015-RMA16-RMA07


© Analysys Mason Limited 2017 The cloud-based management of the packet, optical and fixed access networks

3. The cloud-based management of the packet, optical

and fixed access networks

3.1 Drivers and benefits

The application of SDN to packet/optical transport networks (all technologies across layer 0–3 between DWDM,

optical and IP/MPLS) and fixed access networks is still nascent. It is progressing gradually because these are highly

mature, stable and complex networks with multiple layers, protocols and services. However, with the launch of 5G

networks on the horizon, CSPs anticipate rapid growth in bandwidth capacity demands on mobile backhaul and fixed

next-generation access (NGA) networks. They also face growing competitive pressure to provide more dynamic

enterprise connectivity services to support enterprise cloud access and IoT. These factors are spurring CSPs to think

about holistic and automated operations across all their fixed network technologies. End-to-end, SDN-driven

management and control of fixed networks promise the following benefits:

• reduced capex through improved network utilisation with multi-layer traffic steering and intelligent path

optimisation, based on analytics and emerging AI technologies

— this reduces the need for overprovisioning and deferring expensive capacity upgrades

• reduced opex due to the simplification and automation of manual and time-consuming operations with on-

line and real-time data/inventory-enabled planning, optimisation, provisioning and assurance

• new revenue generation with dynamic, on-demand network services across optical, metro and backbone

networks including elastic bandwidth, NaaS and data-centre interconnectivity, underpinned by policy-based

control, improved security and traffic prioritisation

• market differentiation with enhanced enterprise user experience through customer self-service ordering and

management, coupled with rapid service delivery.

3.2 An ideal multi-layer SDN platform will pave the way to DNOP

To realise the benefits of cloud-based packet/optical transport and fixed access networks, CSPs will need a ‘multi-

layer SDN control’ platform, as illustrated in Figure 5. Such a platform rationalises network management silos by

providing unified, end-to-end and centralised management and control of multiple fixed network domains. The multi-

layer control platform will need to extend incumbent vendor and layer-specific E/NMSs and network control planes

and domains (such as IP/MPLS, optical, microwave) and deliver a high level of programmability while allowing

CSPs to utilise their existing investments without ripping and replacing existing infrastructure.



Figure 5: High-level functional blocks of multi-layer SDN control platform [Source: Analysys Mason, 2017]

Multi-layer SDN control and management platforms should provide at least the key functionalities and features

detailed in Figure 6 below.

Figure 6: Functional capabilities of multi-layer SDN control and management platforms [Source: Analysys Mason,

2017]

Solution requirement Description

Full network lifecycle

management

End-to-end connectivity management as well as OSS-like functionalities such as network

planning (supporting real-time capacity management, ‘what-if’ analysis and ‘just-in-time’

optimisation capabilities), network management, monitoring, assurance and provisioning

through a unified system and operations graphical user interface (GUI)

Multi-layer control and

optimisation

Centralised, real-time, multi-layer network topology view and ability to control and interwork

all layers and services with resource pooling, advanced monitoring and predictive analytics

capabilities

Traffic

steering/engineering

Near real-time or real-time computation and implementation of traffic paths across multiple

network layers to improve utilisation, avoid congestion and provide differentiated SLAs

Flow-based networking Application-driven, flow-based networking, e.g. segment routing, without the need for new

protocols such as OpenFlow

Open NBIs

CD NO

Control plane

abstraction/

extraction

Policy and rules engine

Real-time analytics,

assurance and security

Plan, provision and

optimise

Visualisation and

path computation

Multi-vendor adapters/integration

YANG/NETCONF, T-API, PCEP, SNMP/CLI, etc.

Real-time inventory

Network service

models

Vendor 1

E/NMS

Multi-vendor

configuration

Self-service portals Applications

Vendor 1 SDN

controller

Domain 1

…

Domain n

Vendor 2

E/NMS

Vendor 2 SDN

controller

Domain 2

E2E multi-layer/multi-domain management and control

Management functions

Control functions

Connectivity

VNFM/VIM



Solution requirement Description

Automated re-routing and

remediation

Application/service/network-aware rapid, policy-based traffic re-rerouting, e.g. diversion from a

network node that is congested or under attack (D-DOS) to a healthier/higher-performance part of

the network. This will eventually be supported by machine learning and AI

Modularity Plug-and-play functionalities and compatibility, and integration with a coherent set of third-

party and open-source applications and components such as multi-vendor SDN controllers

(ODL, ONOS), path computation and optimisation engines

Multi-vendor integration

and configuration

Automated control and configuration of hybrid, multi-vendor physical and virtual WAN

elements and their management systems, potentially with a hierarchical abstraction

architecture

Orchestration/scale Vendor-agnostic, seamless northbound integration with traditional OSS and new MANO layer

through open APIs for complete orchestration and management of hybrid network

Open-source based The platform should be built on a substrate of open-source technologies

Standards and model-

based

Support for industry standard data modelling languages, protocols and APIs, both new and

old, such as YANG/NETCONF, PCEP, T-API and SNMP/CLI11

Cloud-native architecture Set of common and shared micro-services/functions; can accommodate rapid onboarding of

new functions to meet future requirements. Support for all types of cloud deployment

models: private, software as a service (SaaS) and hybrid

Self-service provisioning Self-service portal capabilities to enable CSP customers to self-order, provision, monitor and

change/terminate services

Traditionally, network lifecycle management functions such as network planning, assurance and inventory reside in

the offline OSS domain, while real-time control is embedded in the network. However, as described above, multi-

layer control platforms should converge network management and control. Initially, these platforms will integrate

with OSS and orchestration. This will enable the multi-layer control platforms to support some of the key OSS

functionalities by providing a real-time view of network topology and elements which can fulfil the role of ‘dynamic’

inventory for planning and provisioning processes and performing network analytics to support assurance. In the long

term, most of these network management functions will move to multi-layer control platforms. Combined with new

network and service orchestration technologies, the multi-layer control platform will pave the way for the migration

from today’s heavy, non-real-time and siloed OSS to a new and more agile operational platform for cloud-based

networks, and eventually, to the DNOP. This will allow CSPs to re-engineer, streamline and coalesce today’s

disparate and manual network lifecycle management tasks to implement policy/rule-based, closed-loop automated

network operations. Possibly in the future, it would also enable CSPs to achieve AI-driven, more

intelligent/autonomous networks that minimise, if not eliminate, human intervention from day-to-day processes to

achieve maximum operational efficiency.

11 YANG: Yet Another Next-Generation Protocol; NETCONF: The Network Configuration Protocol; PCEP: The Path Computation

Element Communication Protocol; T-API: SDN Transport API; SNMP: Simple Network Management Protocol; CLI: Command

Line Interface.


© Analysys Mason Limited 2017 About Huawei’s NCE in cloud-based networks

4. About Huawei’s NCE in cloud-based networks

4.1 Huawei’s primary goal is achieving network intelligence

Huawei’s focus is on achieving a network that is simple, yet intelligent. It proposes three steps to building network

intelligence: moving from automatic to adaptive, and ultimately to autonomous networks (see Figure 7).

Figure 7: The three essentials of building network intelligence [Source: Huawei, 2017]

These three network-building essentials can be described as follows:

• ‘Automatic’ network – the ability to automatically provision services via intent-based APIs and automate

network deployment and maintenance through full lifecycle management. By integrating network managers

and controllers, the automatic network simplifies the connections and operations of new SDN networks and

traditional networks, which will enable one-click services and end-to-end automatic service provisioning.

• ‘Adaptive’ network – this type of network adds highly proactive analysers to existing automation to

collect real-time network data and assess network status. The adaptive network is able to automatically

generate and optimise policies based on service and network SLAs, enabling networks to evolve from open-

loop configuration to closed-loop optimisation.

• ‘Autonomous’ network – at this stage, analysers are enhanced by introducing AI and machine-learning

algorithms to enable network self-learning. Networks that follow programmed, static policies will evolve to

a state where they learn about and absorb dynamic policies on their own, thus achieving network autonomy.

4.2 NCE is a cloud-based, autonomous network management platform

Huawei NCE (Network Cloud Engine) is a multi-layer, multi-domain SDN solution for CSPs that integrates the

management, control and analysis of next-generation fixed access, packet and optical networks. NCE is

positioned as a ‘brain’ that provides the intelligence for Huawei’s future cloud-based networks. It automates

connectivity in a hybrid network environment by managing and controlling physical and virtual devices in IP,

MPLS, optical, microwave and fibre access networks. It carries out intelligent network operations and

maintenance, and guarantees network connectivity SLAs using visualisation, near real-time network telemetry



and big data analytics. NCE integrates with a service orchestrator to realize end-to-end orchestration of services

across fixed networks, wireless networks, core networks and cloud platforms. NCE exposes network capabilities

through APIs, allowing CSPs to manage multiple enterprise tenants’ connectivity services and offer an e-

commerce-style service experience. It is built as a cloud native platform and it is an essential part of the process

of migrating to all-cloud architectures and digital operations.

Figure 8: Huawei NCE architectural overview [Source: Huawei, 2017]

Key NCE functions

Automating network connections through metamodels and intelligent routing algorithms

NCE provides unified access to traditional physical networks and virtual SDN/NFV networks. Third-party

domain controllers can be integrated with the architecture through open APIs and physical and virtual resources

can be networked together. NCE supports on-demand enterprise access to fixed access and core connectivity

services including xDSL, PON, Ethernet, IP/MPLS, and OTN. On the bearer network side, on-demand

networking is enabled for IP/MPLS and optical networks that span cities and provinces. Centralised

management, control and scheduling of network resources enables automated service provisioning in minutes,

the flexible adjustment of bandwidth and increased automation of service configuration and fault recovery. An

intelligent routing algorithm supports best cost, lowest latency or shortest paths.

Guaranteeing network connection SLAs: policy-driven closed loop based on big data analysis

NCE can assure the operating status of the network through traditional SNMP and embedded telemetry, which

allows the collection of large amounts of data in microseconds across a wide range of scenarios. Networks and

the status of network connections are analysed in near real time using a big data platform, which means that

SLAs for network connections are visual, quantifiable and enforceable. The ECA (Event Condition Action)

policy engine proactively detects risk by means of enhanced policy conflict detection, auditing and simulation,

and automatically triggers the optimisation of networks and connections. The solution provides service

guarantees to ensure service experience for customers, and reduces service interruptions and degradations

through risk prediction and intelligent scheduling.

Automating network operations and maintenance throughout the lifecycle

NCE integrates network management, control and analysis functions through a framework for service-based

component integration. Unlike traditional operations and maintenance (O&M), which required multiple systems,

the NCE is designed around the concept of a single, integrated O&M system that covers the complete lifecycle,



from network planning and deployment, to service provisioning, testing and verification, and finally to service

guarantees and network maintenance. NCE includes a Web-based user operations centre and dashboard to

reduce the need for manual operations in O&M scenarios. All simple, repetitive and compute-intensive tasks

will gradually be automated to reduce errors and increase efficiency. NCE includes an Apps Centre which will

be loaded with frequently used modules to assist with new applications creation. If this is not sufficient, it also

permits individuals to add new modules developed locally to the Apps Centre’s database.

The platform’s open APIs facilitates integration with CSPs’ open O&M cloud platforms to ensure a consistent user

experience and to meet the operational challenges of complex, multi-vendor and multi-technology environments.

Moving toward network intelligence through AI algorithms and big data

NCE is built using an open architecture. Its policy engine and network analysis capabilities use multiple sources

of data and AI algorithms to drill down to root causes, eliminating issues as they occur, and predicting future

issues. Network data is continuously mined to build knowledge and machine-learning used to detect anomalies.

NCE will eventually bring about an intelligent, autonomous network that can operate unattended.

4.3 NCE aims to provide a smooth evolution to autonomous networks

NCE is being developed in response to CSP engagements around the world and small-scale trial projects

conducted with these CSPs over the past three years. In the short term, the solution is focused on helping CSPs

automate enterprise connectivity services across physical, fixed networks with e-commerce-style service

experience and guaranteed SLAs. In the long term, the NCE will focus on network reshaping for 5G, IoT, high

quality broadband and cloud-based networks. It aims to be a platform for network autonomy and support CSPs

in making their operations digital and intelligent over the next 10 years.

NCE offers CSPs a variety of functions to facilitate their migration to cloud-based networks, with the ability to

decompose network silos at their own speed of adoption as it supports the on-demand turn-up of modular

functions and features. The network control component of the NCE incorporates and extends the capabilities of

Huawei’s Agile Controller, enabling an easier upgrade path for existing Agile Controller deployments to the

new NCE platform. A lightweight version of the Agile Controller will continue to be available for small and

medium-sized enterprise customers. A wide variety of business models and agile products can be combined to

meet the needs of customers in different countries and regions operating networks of different scales and in

different scenarios. As a platinum ONAP member, Huawei has actively contributed to and learned from the

development, systems integration and technical architecture of that project. With the benefit of this experience,

Huawei has designed its NCE architecture to enable efficient integration with the proprietary systems of third

parties and CSPs.


© Analysys Mason Limited 2017 Conclusion

5. Conclusion

Cloud-based, highly ‘software-ised’ networks, coupled with a new operational paradigm that is based on a high

degree of autonomy, should be the end goal for CSPs that want to become DSPs and assert themselves in the new

digital service value chains. CSPs will need to migrate from the multiple OSS/EMS/NMS silos they operate in their

traditional networks towards horizontalised, end-to-end network control and management platforms. The DNOP

vision discussed in this paper is the eventual goal for their future, hyper-converged SDN/NFV-based networks.

This transformation should be executed by means of a gradual migration strategy. CSPs should leverage SDN

and NFV technologies in combination to automate various network domains and services step by step, based on

their business objectives and operational needs. Then they should look at ways of stitching discrete platforms

together over time. Many CSPs have started with single domains, for example, mobile core network

virtualisation for capex/opex savings and service agility and/or SD-WAN in combination with vCPE in the fixed

edge network to deliver more dynamic enterprise services to grow/protect revenue. They are also targeting

multi-layer, programmable control of the packet/optical transport network (IP, optical, microwave) to gain

significant cost savings thanks to higher network utilisation and automated operations, and new revenue

opportunities through on-demand bandwidth, NaaS and data-centre interconnectivity services. However,

introducing an SDN-based approach across multiple network domains will be a slow and disruptive process.

CSPs will need open, modular and cloud-native multi-layer control platforms that can provide the following key

capabilities to help them make this transition:

• extend existing vendor and layer-specific E/NMSs, control planes and routing protocols for more centralised

and unified management and control of multiple transport network layers that are traditionally managed

separately (i.e. IP and optical)

• provide end-to-end, automated connectivity management and control in multi-vendor WAN environments

with real-time visibility of network topology

• perform full lifecycle management, including some of the existing OSS functions (planning, optimisation,

provisioning and assurance) to support the move towards future DNOPs.

Huawei’s NCE platform can be placed within Analysys Mason’s SDN multi-layer control platform category.

Based on the key requirements outlined in this paper, we see the following characteristics of the NCE as its

main strengths:

• NCE’s proposed multi-layer/domain capabilities are more extensive than similar competing solutions from

other vendors, which focus on specific network domains and/or layers.

• Its modularity is a good fit for a step-by-step migration strategy towards cloud-based networks. Its end goal

of bringing these specific modules together as micro-services in a horizontalised, cloud-native platform

resonates with our DNOP vision.

• Its hierarchical architecture is supported by a variety of new and old data models. This is a sound approach

as it allows abstraction of domain-specific connectivity and services while ensuring end-to-end full lifecycle

automation.

We provide the following recommendations to Huawei:

• NCE’s domain scope does not include mobile access (RAN) and evolved packet core network management,

for which it is developing the Mobile Cloud Engine (MCE). Huawei should ensure early alignment between


© Analysys Mason Limited 2017 Conclusion

these two platforms (NCE and MCE), despite the organisational complexity this will entail, to reduce

confusion and potentially converge them as a single 5G/DNOP solution in the long term.

• Huawei promotes multi-vendor and openness in the NCE platform, but it currently appears focussed on its

own solutions. This is an approach similar to what we see in most NEP SDN strategies; multi-vendor

interoperability does not become a development priority until the vendors are ready to support their own

solutions first. It will be important for Huawei to demonstrate support for third-party solutions across all

NCE domains if the company wants to be distinctive.

• Cloud-based networks will introduce new security challenges, and SDN control will be at the core of new

network security mechanisms. NCE should provide industry-leading security capabilities and emphasise

security in its messaging.


© Analysys Mason Limited 2017 About the authors

About the authors

Gorkem Yigit (Senior Analyst) is the lead analyst for the Service Delivery Platforms

programme and a contributor to the Software-Controlled Networking and Network

Orchestration programmes, focusing on producing market share, forecast and research

collateral. He started his career in the telecoms industry with a graduate role at a leading

telecoms operator, before joining Analysys Mason in late 2013. He has published research on

NFV/SDN services business cases, identity management in the digital economy, and has been

a key part of major consulting projects including Telco Cloud Index and IPTV/OTT

procurement. He holds a cum laude MSc degree in Economics and Management of Innovation

and Technology from Bocconi University (Milan, Italy).

Caroline Chappell (Principal Analyst) is the lead analyst for Analysys Mason’s Software-

Controlled Networking research programme. Her research focuses on service provider

adoption of cloud and the application of cloud technologies to fixed and mobile networks. She

is a leading exponent of SDN and NFV and the potential that these technologies have to

enhance business agility and enable new revenue opportunities for service providers. Caroline

investigates key cloud and network virtualisation challenges, and helps telecoms customers to

devise strategies that mitigate the disruptive effects of cloud and support a smooth transition to

the era of software-controlled networks. Caroline has over 25 years’ experience as a telecoms

analyst and consultant.

This white paper was commissioned by Huawei. Analysys Mason does not endorse any of the vendor’s products

or services.

12

12

Published by Analysys Mason Limited • Bush House • North West Wing • Aldwych • London • WC2B 4PJ • UK

Tel: +44 (0)20 7395 9000 • Email: [email protected] • www.analysysmason.com/research

Registered in England No. 5177472

© Analysys Mason Limited 2017

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means – electronic,

mechanical, photocopying, recording or otherwise – without the prior written permission of the publisher.

Figures and projections contained in this report are based on publicly available information only and are produced by the Research Division of Analysys Mason

Limited independently of any client-specific work within Analysys Mason Limited. The opinions expressed are those of the stated authors only.

Analysys Mason Limited recognises that many terms appearing in this report are proprietary; all such trademarks are acknowledged and every effort has been

made to indicate them by the normal UK publishing practice of capitalisation. However, the presence of a term, in whatever form, does not affect its legal status

as a trademark.

Analysys Mason Limited maintains that all reasonable care and skill have been used in the compilation of this publication. However, Analysys Mason Limited

shall not be under any liability for loss or damage (including consequential loss) whatsoever or howsoever arising as a result of the use of this publication by the

customer, his servants, agents or any third party.


© Analysys Mason Limited 2017 About Analysys Mason

About Analysys Mason

Analysys Mason is a trusted adviser on telecoms, media and technology. We work with our clients, including

communications service providers (CSPs), regulators and end users to:

• design winning strategies that deliver measurable results

• make informed decisions based on market intelligence and analytical rigour

• develop innovative propositions to gain competitive advantage.

We have around 232 staff in 14 offices and are respected worldwide for exceptional quality of work,

independence and flexibility in responding to client needs. For over 30 years, we have been helping clients in

more than 100 countries to maximise their opportunities.

Consulting

• We deliver tangible benefits to clients across the telecoms industry:

• communications and digital service providers, vendors, financial and strategic investors, private equity

and infrastructure funds, governments, regulators, broadcasters, and service and content providers.

• Our sector specialists understand the distinct

local challenges facing clients, in addition to the

wider effects of global forces.

• We are future-focused and help clients

understand the challenges and opportunities that

new technology brings.

Research

• Our dedicated team of analysts track and

forecast the different services accessed by

consumers and enterprises.

• We offer detailed insight into the software,

infrastructure and technology delivering those

services.

• Clients benefit from regular and timely intelligence, and direct access to analysts.

For more information, please visit www.analysysmason.com/research.

http://www.analysysmason.com/research



Research from Analysys Mason

We provide dedicated coverage of developments in the telecoms, media and technology (TMT) sectors,

through a range of research programmes that focus on different services and regions of the world

The division consists of a specialised team of analysts, who provide dedicated coverage of TMT issues and

trends. Our experts understand not only the complexities of the TMT sectors, but the unique challenges of

companies, regulators and other stakeholders operating in such a dynamic industry.

Our subscription research programmes cover the following key areas.

Each subscription programme provides a combination of quantitative deliverables, including access to more

than 3 million consumer and industry data points, as well as research articles and reports on emerging trends

drawn from our library of research and consulting work.

Our custom research service offers in-depth, tailored analysis that addresses specific issues to meet your

exact requirements

Alongside our standardised suite of research programmes, Analysys Mason’s Custom Research team undertakes

specialised, bespoke research projects for clients. The dedicated team offers tailored investigations and answers

complex questions on markets, competitors and services with customised industry intelligence and insights.

For more information about our research services, please visit www.analysysmason.com/research.

http://www.analysysmason.com/research



Consulting from Analysys Mason

For over 30 years, our consultants have been bringing the benefits of applied intelligence to enable

clients around the world to make the most of their opportunities

Our clients in the telecoms, media and technology (TMT) sectors operate in dynamic markets where change is

constant. We help shape their understanding of the future so they can thrive in these demanding conditions. To

do that, we have developed rigorous methodologies that deliver real results for clients around the world.

Our focus is exclusively on TMT. We advise clients on regulatory matters, help shape spectrum policy and

develop spectrum strategy, support multi-billion-dollar investments, advise on operational performance and

develop new business strategies. Such projects result in a depth of knowledge and a range of expertise that sets

us apart.

We look beyond the obvious to understand a situation from a client’s perspective. Most importantly, we never

forget that the point of consultancy is to provide appropriate and practical solutions. We help clients solve their

most pressing problems, enabling them to go farther, faster and achieve their commercial objectives.

For more information about our consulting services, please visit www.analysysmason.com/consulting.

http://www.analysysmason.com/consulting

Date post:	19-Mar-2018
Category:	Documents
Upload:	phamthuy
View:	213 times
Download:	1 times

Preparing the packet, optical and fixed access networks ... · PDF fileAbout Huawei’s...

Documents