preparing the packet, optical and fixed ......this white paper outlines the key drivers and...
TRANSCRIPT
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
Gorkem Yigit and Caroline Chappell
DECEMBER 2017
WHITE PAPER
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | i
© Analysys Mason Limited 2017 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
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 2
© 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.
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 3
© 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/.
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 4
© Analysys Mason Limited 2017 Starting the journey to a cloud-based network
• 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
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 5
© Analysys Mason Limited 2017 Starting the journey to a cloud-based network
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.
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 6
© Analysys Mason Limited 2017 Starting the journey to a cloud-based network
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
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 7
© Analysys Mason Limited 2017 Starting the journey to a cloud-based network
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.
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 8
© Analysys Mason Limited 2017 Starting the journey to a cloud-based network
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
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 9
© 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.
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 10
© Analysys Mason Limited 2017 The cloud-based management of the packet, optical and fixed access networks
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
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 11
© Analysys Mason Limited 2017 The cloud-based management of the packet, optical and fixed access networks
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.
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 12
© 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
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 13
© Analysys Mason Limited 2017 About Huawei’s NCE in cloud-based networks
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 milliseconds 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,
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 14
© Analysys Mason Limited 2017 About Huawei’s NCE in cloud-based networks
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.
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 15
© 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
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 16
© 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.
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 17
© 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.
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 18
© 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.
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 19
© Analysys Mason Limited 2017 About Analysys Mason
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.
Preparing the packet, optical and fixed access networks for the digital era: requirements for a cloud network platform | 20
© Analysys Mason Limited 2017 About Analysys Mason
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.