From pipelines to clouds Etisalat’s playbook€¦ · services require special capabilities that today’s network infrastructure cannot provide. This is where telco’s emerging

From pipelines to cloudsEtisalat’s playbook

03

From pipelines to clouds | Etisalat’s playbook

Contents

04Cloudification journey

31Contributors’ point of view

34About us

35Further details

Etisalat’s considerations for the digital transformationThe telecommunication industry aroundthe world is experiencing an exponentialgrowth in the number of devicesconnected that ultimately impacts theinfrastructure that needs to accommodatethis data traffic growth. Hence, telcos areconstantly investing in their networkinfrastructure so they can meet the marketdemand. However, this could lead to aprofitability trap, since these investmentsdo not necessarily mean revenue growth.Additionally, the strong competition amongOTT providers is also impacting telcos’revenues, which are stagnating throughoutthe globe or not increasing at the samerate as the network’s investments [1].

Furthermore, customers’ behaviour ischanging. Nowadays, they seek self-servicing, multi-channel offers, anddemand to be always connected to theInternet. This situation is pushingtelecommunications operators to evolvetheir business models from a typicalcommunications service provider (CSP) toa digital service provider (DSP) in whichthey offer innovative value-added services(VAS) and improve quality of experience(QoE) while optimising their networkinfrastructure.

The adoption of a software-definednetwork (SDN) and network functionsvirtualization (NFV) are crucial parts of atelco’s transformational journey. Thesetechnologies are enablers for telcos to staycompetitive in today’s market by reducingthe Time-to-Market (TTM), increasingrevenue, reducing cost of service andimproving asset efficiency. For Etisalat, ourmain drivers to embark on this journey arelisted as follows in Figure 2.

04


Cloudification journey

SIM connections *excluding M2M

M

C 2.2%

0.2%

M

CLAC

North America

4%

0.9%

M

C 1.7%

0.7%

M

C 5%

2.9%

M

C 3.2%

0.9%

M

C 6.1%

2.8%

Europe

MENA

AfricaAsia Pacific

Figure 1: Anual variation (CAGR) forecast for mobile networks from 2015 to 2020

Mobile revenueM C

Figure 2: Main drivers for Etisalat’s cloudification journey

Product agility (e.g. quicker product development,shorter TTM, faster testing, Data-driven decisions)

Cost savings (e.g. pooling, reusability, modularity, new TCO models, more options)

Self care (e.g. digital experience, self-provision & troubleshooting, mass customizations)

Service agility (e.g. real-time, on-demand, faster activation and cessation)

Auto healing (e.g. auto-assurance, artificial intelligence, experiential network intelligence)

1st

2nd

3rd

4th

5th

New services and user experiencetechnological trendsApart from the market drivers to embarkon a transformational journey, telcos willneed to unlock and explore new sources of revenue from digital services fulfillingthe customers’ digital experienceexpectations. Most of these value-addedservices require special capabilities thattoday’s network infrastructure cannotprovide. This is where telco’s emergingtechnologies combined (SDN, NFV, cloudcomputing, 5G, MEC, IoT) will play animportant role in enabling telcos and theircustomers to develop new products to beoffered into the market as illustrated inFigure 3 (non-exhaustive).

Not only that, but we see that emergingtelecom technologies will effectively unlockchoices not available today to telcos whendealing with OTTs and other digitally-transformed organizations: to compete orto collaborate on a comparative footing.

The use cases illustrated in Figure 3 arepotential business and new source ofrevenue for telcos. To exemplify, we canlook at the drive assistance use case,which leverages technology to overcomehuman limitations. Let us assume that aperson is driving at 100Km/h and suddenlya car crashes in front of their vehicle. Thereaction to break happens in 1 secondand, therefore, the car would move at least 27 meters more, enough to cause an accident.

A study conducted by Deloitte in the USAmarket predicts that safety will be one ofthe priorities of citizens in the future andthat by 2020 a sixth of the cars will beequipped with automatic emergencybreaks (AEB) that will contribute to a 16%reduction in motor vehicle deaths. The AEBis enabled by connected vehiclestechnologies (V2X – vehicle to everything)that requires either vehicle-based sensorsor wifi /mobile (4.5/5G) communication [2].The 5G mobile networks will overcomemost of the sensors’ limitations (forinstance, range) [3] giving an opportunityfor telcos to explore this market segment.Similarly to V2X, which requires a mobilenetwork with ultra low latency, other usecases will also rely on special networkparametrization. The IoT services, forinstance, applied in smart cities, new videoservices (VR – virtual reality) and cloud B2Bservices (IaaS, SaaS) are some examples tobe explored. Tapping into this blue oceanis neither economically feasible nortechnically possible without telcosvirtualising their infrastructure and digitallytransforming in the process.

05


Figure 3: Use cases for value-added services with special network requirements

Source: Deloitte TMT predictions, 2017

Ultra-reliablecommunication

• eHealth• Remote surgery/ remote object control• Natural disaster

• Sensor networks• Smart wearables (clothes)• Mobile video surveillance• Drones control• Critical control of remote devices

Massive IoT

Ultra-low latency

• Drive assistance, Autonomous driving• Smart traffic control, Smart infrastructure• Tactile control

Broadband access everywhere

• +50 Mbps everywhere• Dense urban society• Hd video/photo sharing in stadium/open air gathering• Smart Office On-the-go

High user mobility

• Moving hotspot• High speed train• Remote computing• Aircrafts

Broadcast-like services

• Live-TV at scale• Broadcast-like services-local, regional, and national• On-demand anything

Muscular reaction

Figure 4: Order of magnitude of human reaction times

1S Auditory reaction

100mS

Visual reaction 10mS

Sense reaction 1mS

06


In order to achieve the goals previouslymentioned, telcos must launch a digitaltransformational programme which willimpact multiple dimensions in theirorganizations. Based on TM Forum’s digitalframework [4], we believe that the gear tomove the organization towards its digitalvision is composed of five dimensions(represented in the core of the gear inFigure 5), encompassed by agile, recursiveiterations that will keep the core in motionacross the journey. The digitaltransformation will be challenging fortelcos that will need to address multipleconcerns related to their journey. Wesummarise all of them by grouping theminto three main challenges as illustratedbelow. The “gear” is the mechanism tokeep moving forward and also keep thefive dimensions developing andsynchronised throughout the journey.

At least three strong shifts will be faced inthis important journey: 1) Internalcannibalization concerns, 2) Existinginvestments and skillsets, and 3)Traditional telecom vendors concerns.Although not directly tackled by this whitepaper, we believe the telco’s over-archingtransformational program mustcontinuously alleviate them openly andreasonably as they embark on theCloudification Journey.

It is important to highlight that thetraditional, siloed operating model oftelcos today has to gradually but surely dieout, since it is probably the biggest softbottleneck there is to this journey. Agility isnot just speed—it is a philosophy withprinciples.

Even though all dimensions are importantand provide a solid foundation for eachother, for this specific paper Etisalatintends to deep dive into the infrastructureand systems dimension to explore whatwill be called the ‘cloudification journey.

The digital transformation and itschallenges to be addressed

Strategy

Services &

customers Infrastructure

& systems

Operations &

processes

Culture,

people &

organization

Assess

Imagine

Design

Exec

ute

Assess: Pinpoint thetelco in the overall journey and go over its maturity levels and readiness to progress

Imagine: Explore opportunities & set the best path towards the target scenario

Execute: Deliver functional outputs to generate value to the company until reaching the target scenario

Design: Plan in detail how the selected path will be executed

First challenge: Define the origin, readiness to move, destination and the drivers to ramp up on the journey

Second challenge: Pinpoint the progress throughout the journey

Time Vector

Altitude

Third challenge:

: keep the core & the pace throughout th

e journey

100%

0%

Targ

et s

cena

rio

fulfi

lmen

t

Average speed vector

Figure 5: Digital transformation framework(Based on TMFORUM)

Figure 6: Illustration of the digital transformation journey and itschallenges

07


Four main ‘ages’ are depicted to breakdown Etisalat’s cloudification journey as along-term transformational programme asdescribed below.

The cloudification journey brokendown into ages

Agile development and deployment in a DevOps culture,

enabled by software

Lower TCO with new sources ofrevenue and improved QoE

Flexible infrastructure, servicescalability and zero-touch provising

and resolution

Digital infrastructure

Orchestration

Automated control

Today Tomorrow

Network Compute Storage

Manual control

Manual control

Manual control

Dedicated hardware makes service scalability costly and inflexible

Time-consuming development, based on a waterfall methodology

Human-driven procurement and operations limit new services

development

Being

Being

Business, operating and customer models are leveraged by digital technologies

Becoming

Becoming

Leveraging digital technologies to create new capabilities

Doing

Doing

Leveraging digital technologies to extend capabilities

Exploring

ExploringGain traction on digital technologies Still leveraging classical technologies

Etisalat’s cloudification ages*

Time vector

“Altitude” vector

100%

0%

Targ

et s

cena

rio

fulfi

lmen

t

*Based on Deloitte’s digital maturity assessment framework

Figure 7: The cloudification journey’s ages

08


Etisalat envisions the target scenario of theinfrastructure to be simpler, more agile,more flexible and more software-based,with control centralised and intelligencepushed to the edge. Just as fibre-optics areindifferent to the kind of traffic or servicerunning over it, Etisalat is evolving towardsa network that is similarly transparent,enabling Etisalat to flexibly and freelylaunch any future service on thisinfrastructure. Service orchestration seemsto be an extremely important componentto achieve these goals. We believe that it isparamount that orchestration standardisesfaster and evolves more quickly to areliably mature level in the coming years.One of Etisalat’s initiatives is building atelco ‘app marketplace’, and wishes forsimilar marketplaces to rise in the industry,akin to the Apple Store or Google Play butfor telecom services. This ‘cloud factory’ or‘telco app factory’ can host new telcoapplications (software-based) that areeasily upgradable and zero-touchprovisioned in the new network by the

service orchestrator to provide newservices to consumers, businesses and thetelco itself (e.g. for its OpCos). For Etisalatto enable this cloud factory, anotherimportant component is the NFVinfrastructure. This layer must become acommodity (universal NFVI), shifting theresponsibility of service resiliency fromhardware to software (e.g. infrastructure ascode), and from the network to the telco’sapplications. This, Etisalat learned, is alsochanging how networks will be designedand what redundancies will beimplemented, and how upcoming telcoservices are built to be more empoweredfor more autonomous operation, ratherthan relying heavily on classical networking(e.g. auto capacity management and autohealing). Overall, the NFV infrastructuremust be aligned and planned well (with themaster plans) for increasing butharmonised distribution towards the edge(which may include other OpCos andgeographies) to ultimately displace andreplace legacy services. An exception to

this smooth distribution towards the edgeis the CPE, the virtualization of which cancome very early in the journey, becoming amicro NFVI in the customer premises, andunlocking new opportunities for multi-service offerings with very low time-to-market (TTM).

A glimpse of the target scenario(Infrastructure & systems dimension)

Service deployment better experience

Dynamic resource allocation zero-touch provisioning

Service orchestrator

NFV orchestrator SDN controller(s)

Digital portal

Customers (B2C/B2B)

Service self purchase

NFVI (NFV infrastructure)

Telco app factory (services & features)

Consortium composed of vendors, telcos and co-

developers to launch new applications and features

NG-OSS/BSS

Big data CRM CEM Inventory

Verticals OTTs

telcos

PNFs

API integrations Legend:

The vision for our futurearchitecture is a fluidmicro-service-basedinfrastructure, leveragingmodern APIs, andconverging on a platformthat will automate theend-to-end servicelifecycle. The serviceorchestrator, big dataand AI are at the heart ofthis future architecture.

Figure 8: High-level envisioned infrastructure and system scenario

09


Telcos will have to develop a plan toimplement a network datacentre layer thatwill be positioned in areas/regions closerto potential customers keen on consumingdigital value-added services. Thedatacentres will be composed of COTShardware managed and operated by acloud operating system, and switches (orwhite boxes) controlled by an SDNcontroller. For instance, OpenStack,overseen by the OpenStack Foundation [5],provides a suite of open-source tools tobuild and manage cloud computingplatforms. The virtualised functions andSDN controllers should be deployed as perthe telco’s roadmap, hosted at the networkdatacentre layer. Complemented by othertechnologies and reference solutions (likeCORD and MEC), they will be the enablersto deploy the use cases that require lowerlatency and higher throughputs.

Additionally, IT workloads, contentconsumed by customers and immediateprocessing needs (e.g. edge analytics) canbe served by the edge networkdatacentres, avoiding data unnecessarilyclogging the core, especially in peak hours,and reducing latency and jitter.

It is important for telcos to decide whichparts of their existing infrastructure (e.g.DWDM and GPON) can be leveraged.Moreover, in areas and markets ofsignificant new growth, this networkdatacentre placement/connectivity and themaster planning of sites/exchanges andcable routes must be aligned, identifyingany changes required to comply with thefuture services requirements and buildingsolid business cases for the newdistributed digital infrastructure.

Exchanges and their re-architecturebased on NFV/SDN

COTS HW

VRAN VCPE VIMS IT app(SaaS)

Digital SDP

Cloud operating system

Virtualised resources

SDN controller

Illustrative Network Datacentre

Edge DC Regional DC National DC

Typical NW location Access & backhaul Backbone & metro Backbone

Service latency 1ms < L< 10ms 10ms < L< 25ms 25ms < L < 40ms

Typical scenarios Ultra low latency services Real-time services & massive bandwidth

Non-real-time services & massive connections

0km 20km 50km >500km

Regional DC National DC

Enterprise

Home

BS

Edge DC

Metro BackboneX-Haul

Access

Figure 9: NFVI high level architecture

10


To achieve the target scenario, a series ofNFV and SDN use cases should bedeployed according to a roadmap thatneeds to be defined by the telco. Amethodology Etisalat used to develop thetransformational roadmap consists ofplotting in a chart the NFV & SDN functionsor use cases to analyse theirimplementation and operation complexityand the anticipated value. The first axis,representing the complexity to implementand operate, will be composed of multiplefactors weighted and scored. For example,integration levels required, changes onnetwork infrastructure and systems,application readiness level, cloudresources required, impact on currentservices/customers and migration impact(offloading legacy services). The secondaxis, which represents the value expected,will also be composed of multiple weightedfactors, such as enabling new services,

enabling third-party integrations,increasing efficiency and automation,enhancing user experience and TCOreduction. Telcos who want to embark onthis journey can develop a similarapproach adapted according to theirstrategy and operating model. It isimportant to highlight that the roadmapmight be regularly re-assessed andupdated since technology is in constantevolution. It is certainly recommended tostart deploying uses cases with lowercomplexity and higher value, but eachtelco is responsible for deciding its ownroadmap, For instance, to skip virtualisingan obsolete function or service, orlaunching a certain NFV/SDN use case as avertical silo, such as SD-WAN. To illustratethe methodology that Etisalat used, themodel was populated and linked back tothree ages defined in our Cloudificationjourney.

Developing the transformationalroadmap for the cloudification journey

To achieve the target scenario, a series ofNFV and SDN use cases should bedeployed according to a roadmap thatneeds to be defined by the telco. Amethodology Etisalat used to develop thetransformational roadmap consists ofplotting in a chart the NFV & SDN functionsor use cases to analyse theirimplementation and operation complexityand the anticipated value. The first axis,representing the complexity to implementand operate, will be composed of multiplefactors weighted and scored. For example,integration levels required, changes onnetwork infrastructure and systems,application readiness level, cloudresources required, impact on currentservices/customers and migration impact(offloading legacy services). The secondaxis, which represents the value expected,will also be composed of multiple weightedfactors, such as enabling new services,

enabling third-party integrations,increasing efficiency and automation,enhancing user experience and TCOreduction. Telcos who want to embark onthis journey can develop a similarapproach adapted according to theirstrategy and operating model. It isimportant to highlight that the roadmapmight be regularly re-assessed andupdated since technology is in constantevolution. It is certainly recommended tostart deploying uses cases with lowercomplexity and higher value, but eachtelco is responsible for deciding its ownroadmap, For instance, to skip virtualisingan obsolete function or service, orlaunching a certain NFV/SDN use case as avertical silo, such as SD-WAN. To illustratethe methodology that Etisalat used, themodel was populated and linked back tothree ages defined in our Cloudificationjourney.

IInnffrraassttrruuccttuurree MMaattuurriittyy PPaatthh

Multi-VNF NFVI deploymentsCore or concentrated NFVIs

Domain orchestrators

Multi-vendor/multi-VIM metro NFVIsIntegration or merger of silo orchestratorsControl & user plane advanced separation

Cloud-native networkEnd-to-end service orchestration

Fully software-defined networkEdge/fog computing

Silo NFVI deploymentsLegacy infrastructure integration

DO

ING

Low High

DROP

Com

plex

ity

Low

H

igh

vRAN vOLT

vDNS vDHCP

vFirewall

vCGNAT vAAA

vBNG

vIMS

RvCPE

vDPI

vLB WAN SDN

DC-SDN

vHLR

T-SDN

vUDC

Mature

Not Mature

DCI-SDN vCDN

vEPC

vPCRF

SDN-Backbone SDN-Metro

BvCPE

MGW

PSTN

vPE

vSBC

vRR

Value Expected

BECO

MIN

GBE

ING

Illustrative

Graphic 1: NFV/SDN functions maturity assessment

11


After assessing all possible functions oruse cases based on NFV and SDNtechnologies according to their analysedvalue and complexity, combined with theinfrastructure maturity path, telcos will beable to define a feasible plan towards thetarget scenario. The cloudification journeyis broken down into ages representing“waves of deployments”, from low to highcomplexity, that will ultimately composethe roadmap towards being a cloud-nativetelco. Depending on the virtual solutions’maturity and the telco’s virtualizationmaturity, ages can be skipped to speed upthe journey. The ages are briefly describedas follows:

The third age focuses on the NFVI and aligning it with the telco’s sites/exchanges masterplanning mentioned before. Telcos should deploy datacentres (nationally, regionally and inthe metro/edge) according to their roadmap. By extending the digital infrastructure closerto where the customer is, new services that require very low latencies (in both networking,like in 5G, and processing, like in edge AI use cases) can be offered to the market. Tointerconnect all these datacentres and fully abstract the infrastructure layer as one telcocloud, telcos will need to deploy DCI-SDN and DCN-SDN technologies. Going further,telcos could harmonise existing silos by building an intermediary step—implementing anorchestrator of orchestrators for cross-domain services. Including the physical domain isstrongly recommended, since purely virtual services (like mobile VPN) are rare and almostall virtual services will be hybrid. Also in this age, more use cases are added based on thetelco’s roadmap and this new infrastructure (e.g. vCDN, vSBC, vSMSC, vDPI, vUDC, vPCRF,vHLR, vFirewall, etc.).

The fourth age is when the end-to-end service orchestrator becomes mainstream,absorbs all existing silos and fully integrates with other systems. The network andfunctions become cloud-native, with the spread of containers, and perhaps evenserverless computing. The next-generation OSS/BSS (NG-OSS/BSS) and a digitalmarketplace will be fully integrated with the service orchestrator based on APIs.Surrounding them, big data platforms, advanced analytics systems and AI will correlateand learn from multiple sources of data (internal and external, such as social media)yielding vital information to continuously improve the customer experience, eventually inan autonomous way. Based on the new NFVI (or telco cloud), and combined with edge/fogcomputing, digital services can be expanded to reach a wider geographic area and ahigher number of customers to enjoy them. Scaling up the cloudification journey, telcoswill by now be technologically, operationally and culturally mature enough to confidentlydeploy and operate any NFV/SDN use case, including high-complexity ones.

Description of the transformationalroadmap according to the journey ages

1. The exploring age

2. The doing age

3. The becoming age

4. The being age

The first age is composed of NFV and SDNcomponents that can be set up and testedin non-production environments and canbe used to gain traction with thetechnologies. Telcos could start this phaseby assembling a team with anexperimental mind-set and a labenvironment in which they can createprototypical permutations before solutionpilots and live trials. The lab infrastructureand stacks are recommended to be multi-vendor in different combinations to learnabout integration and interoperability inNFV and SDN (perhaps as a precursor tothe Telco’s certification lab). Open sourcesoftware (e.g. OPNFV and ODL/ONOS) andwhite boxes often satisfy this stage.

The second age is composed of a group of NFV and SDN functions and use cases withstrong business cases and/or proven commercial success worldwide. The recommendedgroup for first deployments (usually SD-WAN, BvCPE, vEPC and vIMS) should have highvalue and low complexity that can be translated into quick wins and enhanced businesscases to ‘fund’ the journey. For this phase, the implementation can be done for specificdomains of the network most ripe for virtualization, with their own orchestrator (MANO)and infrastructure (NFVI). This age creates vertical deployments or silos for faster time-to-market, igniting the transformation early, ramping up experience in the new technologiesand encouraging major vendors. With a stronger business case and sustained by the firstwave of deployments, other low complexity and high value functions can be pursued,especially control functions (carrying non-live traffic) or those due for upgrade, expansionor write-off.

12


By design, the adoption of disruptivetechnologies deeply impacts telcos’infrastructures and systems, which mayextend to their products and services.Hence, it is recommended to conductProof of Concepts (PoCs) in order toexplore the technologies, test theirtechnical feasibility and forecast theirimpact. To perform the PoCs, it isnecessary to set up a lab environment withconditions identical to the telco’sproduction network (i.e. with access tostaging test beds and existing networklabs). The PoCs are highly recommendedto adopt standard NFV and SDNarchitectures and protocols (or at leastpopular ones by adoption and size of opensource community), ensuring access to thewidest variety of functions and use cases,as well as interoperability.

Standardization (via standards bodies,open source foundations/communities oreven consortiums) is critical to the successof virtualization. If not joining, telcos shouldat least track the progress and updates ofthese organizations as input to theirvirtualization strategy and roadmaps, aswell as to influence and partner withvendors to embrace these standards, opensource software and open APIs.

Being software-based technologies, NFVand SDN standards are in constantevolution and telcos should always beupdated about new releases that mayaffect their journey. For instance, inOctober 2010, ETSI released a document[6] presenting a set of PoCs and use casesrelated to a NFV and SDN combinedarchitecture, before this specification both

concepts could not be integrated. ETSI andIETF among other entities are trying topush these topics forward. One of theorganizations is TM Forum, which launcheda project called ZOOM [7] that intends todefine the best practices to support boththe technological and business aspects ofthe transformation. Recently, TM Forumestablished a partnership with MEF tocollaborate on the standardization oflifecycle service orchestrator (LSO), anetwork architecture defined by MEF [8] in2016 in which all the network is managedby an end-to-end orchestrator, similar toEtisalat’s approach. Like any new anddisruptive technology, individual non-mature or non-successful NFV/SDNendeavours may bring about skepticismbut should not dictate the fate of the

telco’s transformation. Their evolutionrelies on a combined effort fromstandardization bodies, other telcos,vendors (telecom, IT and consultancies),and open source communities.Collaboration is key, and happens to be acore value of Etisalat.

Even though most vendors base theirsolutions on open-source software, theygenerally modify them extensively, some tothe point where these platforms resembleproprietary solutions more than opensource. Therefore, when selecting thepartner(s), it is important to guarantee theyare adherent to standards, softwarearchitecture principles and the respectivetelco’s strategy in order to achieve thegoals envisioned at the end of the journey.

The exploring ageProof of concepts using industry standards

SDN Bodies

NFV Bodies

Figure 10: Worldwide standardization bodies and foundations landscape

13


Develop your cloudification roadmap, set up a multi-vendor lab environment, assessmulti-vendor interoperability and building in-house experience and expertise

Building a lab enviroment

EMS (1 EMS (1

NFVI

VNF VNF VNF

VIM

VNFM

PNF () PNF (1 PNF

OSS/BSS

Service orchestration NFVO

Dat

a pl

ane

Cont

rol p

lane

SDN Controllers

Catalogs/repositories

Access Nw. SDN controller

WAN SDN controller

DC SDN controller

Other EMS

VNF Vi-Vnfm

Vn-Nf

Or-

Vi

Or-V

nfm

Or-S

dnc

Ca-V

nfm

Sdnc-Nf

Se-Ma

Ve-Vnfm

(towards L2/L3 VNFs)

Figure 11: Lab environment with SDN and NFV combines architecture

14


Finding a cloud-native version of a physicaltelecom appliance or a network function isjust one of the challenges of having acompletely virtualised network. The otherchallenge is to ensure that the underlyinginfrastructure is composed of white boxes,which are, essentially, commercial off-the-shelf servers and commodity bare-metalequipment [9]. One of the biggestpromises of NFV and SDN was preciselythe use of commodity hardware (e.g.servers, switches, CPEs, transceivers) in theinfrastructure, being relatively agnostic tothe software running on them (e.g. multi-vendor hardware and softwarecombinations), which will reduce costs andincrease flexibility as it did for IT and majorweb companies. The white boxes are animportant technological trend innetworking, and one of the first vendors toenable this trend was Intel. The Americancompany developed the Data PlaneDevelopment Kit (DPDK), a set of softwarelibraries and drivers for fast packetprocessing that can run on any processor(e.g. Intel x86, ARM), giving commodityhardware the capacity to handle themajority of the packet loads of currentnetworks [10].

Although there are no ‘standards’ definedfor the design of these types of equipment,many vendors use the designs of the OpenCompute Project (OCP), a project initiatedby Facebook with the purpose of‘economising the datacentre’. For thisinitiative, Facebook already developedseveral systems and equipment such asFBOSS, Wedge and, more recently,Voyager, an optical switch released inNovember 2016. The four datacentres of

Facebook that are 100% powered by OCPhardware and systems were 38% moreenergy efficient to build and 24% lessexpensive to run when compared with thesocial network giant’s previous facilities[11]. Even though Facebook’sachievements concerning the developmentof commodity hardware were impactful,some specialists state that the use of whitebox equipment did not really disrupt datacentres. Instead, it mainly disrupted theserver makers, lowering the prices ofserver components. Either way, telcos alsoreap the benefits [12]. Another project,also with Facebook’s stamp on it, andperhaps more interesting, takes aim attelecom vendors and telcos. With thepurpose of ‘economising the telecominfrastructure’ and modernising how it isbuilt, the Telecom Infra Project (TIP) isworth following.

Nevertheless, the white boxes trend willdeeply change the paradigms of hardwareand software providers (vendors) that willneed to shift from being network-oriented(or product-oriented) organizations toservice-oriented organizations. Vendorshave been incorporating professionalservices into their products, a transitionthat could be explained by an EBIT margin3 to 7 times higher and a continuousrevenue stream [13]. However, thedigitalization (or the transition to beingservice-oriented) of vendors enters a newlevel with NFV/SDN technologies, sincenetwork equipment will no longer demandspecialised or bespoke cutting-edgetechnologies, with the control andmanagement separated and centralised insoftware-based controllers, and functions

virtualised in software. The infrastructurelayer will be based on commodityhardware, abstracted as the telco cloud,and the real service and value will sit withcloud-native applications and services thatvendors and partners will co-developtogether with telcos to generate newrevenues, and yield operational and costefficiencies. A gradual shift will be noticedin focus and spending away from telco’sexpensive hardware towards innovations insoftware.It is important to highlight that theparadigm of a virtualised network asenvisioned by ETSI can only be completelyaccomplished if telcos develop and/orprocure cloud-native VNFs, and run themon a multi-tenant telco cloud, in turnrunning on top of COTS hardware.Traditional telecom vendors areincreasingly adapting and adopting thistrend more quickly, or they would riskshrinking their market shares and beingconfined to legacy pockets of the telecominfrastructure. Corroborating that, weobserve that smaller ‘software-first’telecom vendors, untied to networklegacies and unburdened by entrenchedvalue networks, have started encounteringincreasingly lower barriers of entry in thetypically challenging telecom market.

White boxes and COTS hardware

15


The second age of the cloudificationjourney starts by deploying NFV and SDNuse cases that stand out based on theirvalue or proven success, thanks to the factof being relatively well explored by theindustry (like SD-WAN, vCPE, vEPC andvIMS). Implemented by Etisalat, the usecases have solid business cases and areoffered by various vendors, big and small,old and new, in the market. Therefore, wewill attempt to analyse them in more detail,not only highlighting a couple of their business case figures but alsodepicting how they could be deployed inthis age.

Software-Defined WAN (SD-WAN): Software-defined WAN aims to simplify themanagement of multiple WAN links. It isused to create a software-defined overlaynetwork independent from the carrier orphysical infrastructure, reducing the time-to-market involved, easing the complexityof configuring and managing thecustomer’s WAN and reducing theoperational costs of all parties. For thesereasons, many companies with branchoffices across the globe have becomequickly aware of this technology and areshowing demand in adopting it, sometimeseven without waiting for their local telcosto offer it, by either building it themselveswith SD-WAN vendors or availing of theservice from global telcos. Essentially, anyTelco (or even vendor) may offer SD-WANas a service anywhere on top of even acompeting Telco’s network, with newopportunities to provide cloud-basedvalue-added and OTT services in a self-service fashion, such as applicationperformance accelerators, voice servicesand virtual security functions, andunlocking an easier and faster stream ofrevenue.

In a network architecture that implementsSD-WAN, there are three elements thatmust be present so the Telco can have asolid SD-WAN solution: the thin SDN-enabled CPE, the SD-WAN controller andthe SD-WAN user interface (UI). The SDNCPE provides the capability of using WANconnections with application-based routing(among others). The SD-WAN controller is acentralised software platform that setspolicies and management for all the SDNCPEs. Finally, signifying the self-care aspectof the service, the SD-WAN UI providesservice management and visibility to thecustomer [24], often developed in an agileand DevOps fashion like in smartphoneapp development companies. The moreconservative SD-WAN market prediction isfrom Gartner, which forecasts a growthfrom USD 129 million in 2016 to USD 1.24billion in 2020 [25]. Trying to take theirshare, telcos have been working togetherwith vendors to streamline WANmanagement. In fact, there are alreadymany network operators providing SD-WAN as a managed service and the usecases already deployed have beensuccessful, which makes SD-WAN anattractive service to sell to companies thatwant improved transport options, betternetwork security, intelligent pathwaycontrol and automatic provisioning [26].

For enterprises, there are also significantresiliency benefits, such as substantialreductions in WAN trouble tickets andoutages, as was shown in a NemertesResearch’s study [27]. The enterprises thatparticipated in that study declared thattheir sites still had link problems but thattheir users were not aware of theseproblems because the service was notdisrupted.

Virtual customer premises equipment(vCPE): According to an SDxCentral report[14], SD-WAN and vCPE are the leading usecases in the NFV and SDN market. It is nosurprise that, generally speaking, operatorsdecided to start their journey byvirtualising the CPE, as it is less risky tostart at the edge of the network to mitigatethe impact of outages on customers. Bystarting with the vCPE, it is possible todeploy the technology just for specificclients, allowing the operator to test it in areal environment scenario withoutcompromising all the network.

In the traditional CPE, hardware appliancesare required at the customer premises inorder to provide a network boundarybetween the operator and the customer.The goal of vCPE is to reduce the amountand complexity of equipment needed bydeploying a smaller number of platforms,either locally at the customer’s premise orin a centralised datacentre so the operator(or customer) can remotely instantiate andmanage a variety of functions. Animportant part of this solution is the vCPE(sometimes called the thick vCPE or theuniversal CPE), an often white box CPEacting as a micro-NFVI at the customerpremises.

The business case for vCPE is quiteappealing as shown in a study performedby Analysis Mason, in which it is stated thata Tier 1 operator in a developed marketcan generate USD 1.1bn net present value(NPV) over a five-year period if it is one ofthe first movers in enterprise vCPE(referred to sometimes as BvCPE).Moreover, the study indicated up to 47%BvCPE-related cost savings per site peryear after the migration of all sites [15].The same Analysis Mason study also statesthat residential vCPE migration in a

The doing ageFirst set of deployments with provenbusiness cases

16


developed market can lead to 82% ofvCPE-related cost savings per year perhousehold after the migration of allhouseholds to vCPE (usually referred to asRvCPE). It is important to mention that thisvalue was obtained when considering aTier 1 player with 17 million customerhouseholds over a five-year period. Withsuch appealing business cases, it is normalthat relevant operators around the worldhave been deploying vCPE, both enterpriseand residential. Some examples ofoperators that already implemented vCPEare AT&T, NTT, Telefonica and Verizon [16][17] [18] [19].

Virtual Evolved Packet Core (vEPC): The EPC is the foundation of a mobiletelecommunications operator’s corenetwork through which mobile services arecreated and delivered to customers [21].Recently, with the exponential growth inmobile connections associated with newtypes of services (e.g. high-definitionmobile video streaming) and new ways todeliver those services, telcos have beenchallenged to rethink their networks and tofind new approaches to support suchservices and meet more demandingrequirements that traditional EPC cannotfulfil. One of the solutions that came upwas the virtualization of EPC. Manyrelevant operators around the world arevirtualising their EPC, being motivated bybusiness cases and studies such as theone performed by IDC in which it wasstated that virtualising the entire set of EPCfunctions could enable the mobile networkinfrastructure to operate at a utilizationrate of up to 87%, resulting in OpExsavings of up to 25% [22]. Furthermore, inthe same study, IDC concluded that telcoscan experience a 67% reduction in theTTM to launch new mobile services. Eventhough vEPC solutions are already mature,as has been proven by theannouncements of major telcos likeEtisalat that started to commerciallyvirtualise their EPC [23], there are stillsome key challenges that must beovercome in order for telcos to have asuccessful virtual EPC deployment. First ofall, they need to push their vEPC vendors

(more than the NFVI) to exceed or, at least,maintain the demanding service availabilityand real-time performance requirementsof current networks, guaranteeing that themigration process is transparent for end-users, and not impacting the 99.999%service availability requirements. Asmentioned earlier, service availability in thevirtual network shifts from being mainlyhardware-centric to mainly software-centric resiliency.

Virtual IP Multimedia Subsystem(vIMS): Like the EPC, the IMS has also been facingstringent requirements due to the growthof mobile Internet connections and theofferings of new services such as VoLTEand VoWiFi (Wi-Fi calling). Due to the needfor programmable and scalable corenetworks, capable of answering the needsof different customer types and scenarios,telcos are turning their attention to virtualIMS solutions. One of the attractions of atruly virtualised IMS is the capability tosupport a multi-tenant model. Examplesinclude enabling multinational telcos tohost their operating companies (OpCos) ona centrally managed infrastructure or eventhe ability for a Telco to offer wholesaleIMS services to other telcos. In addition tothe capabilities aforementioned, there arealso the benefits inherent to NFV solutions,such as CapEx and OpEx savings.Concerning this topic, Analysis Masonpublished an article presenting that vIMSfor VoLTE services can lead to 45% CapExsavings and 10% OpEx savings whencompared with a physical on premise IMSfor VoLTE [20].

Other possible deployments:By deploying vDNS, queries for thenameservers are sent to a datacentre,where they are checked in order to provetheir legitimacy and to block malicioustraffic. As it sits in front of the corporation’snameservers, vDNS shields theorganizations by masking the origin IPaddress from the attackers and reroutesthe traffic through the network. Specifically,DDoS attacks are prevented by maskingthe origin IP address with vDNS. Such

attacks are a major concern, according to aDeloitte internal study which predicted athree-fold increase in attack size from2015 to 2017 because of trends such asIoT and higher bandwidth speeds. Otherbenefits can also be generated such asincreased availability and easier scalability.

The vFW is a VNF that enables telcos to offer Firewall as a service to customers. Inaddition, scalability is not a problem sinceit is not necessary to install new instancesof firewall within the network. There aretwo ways a vFW can operate in thenetwork: hypervisor and bridge mode. Inthe bridge mode, the vFW is normallydeployed in a switch intercepting incomingtraffic from other network segments. Byoperating in a hypervisor mode, it isdeployed between the VMs and thevSwitch. This way, it protects the VMs notonly from the incoming Internet traffic butalso from other VMs traffic, working as aeast-west protection.

Deploying a vDHCP as a VNF improvessecurity, as it removes the possibility of asingle point of failure since there is nolonger a dedicated server for DHCP. Also,with vDHCP it is easier to scale the networkand automatize routines. By virtualisingCGNAT, performance is improved andtelcos can quick expand their IPv4networks. Since IPv6 adoption is slowerthan expected, and as more devices aredaily inserted into the network, this VNFs isof great importance to the telco.

Overall, the benefits of these functionsprevious described can be translated incost avoidance. As demand increases andthe servers have a higher utilization rate,there is a better use of infrastructureresources because the need to acquirenew servers can be delayed. When aninstance of a VNF is about to reach itsusage threshold, the server deploysanother VM in opposition to the telcobeing required to buy another hardware tosupport the function.

17


A very important step in the Doing Age isthe implementation of a complete MANO(management & orchestration) stack, aplatform to orchestrate both physical and software resources in a virtualisednetwork. Together with the NFVI (andsometimes with their DC-SDN), they arepopularly and collectively called the telcocloud. Defined by ETSI, the MANO iscomposed of three components: 1) theNFV orchestrator (NFVO), responsible forthe on-boarding of new network servicesand VNF packages, for the lifecyclemanagement of network services, for theglobal resource management and for thevalidation and authorization of NFVIresource requests; 2) the VNF Manager(VNFM), which has the role of guiding thelifecycle management of VNF instancesand for the report of events between theNFVI and the EMSs; and 3) the virtualisedinfrastructure manager (VIM), responsiblefor the control and management of thevirtualised physical resources (NFVI). A veryimportant aspect of MANO integration isrelated to APIs. In order to work properlyand effectively, the MANO must beintegrated with open APIs in the existingsystems. This fact not only affects vendorsthat intend to provide any part (or all parts)of a MANO solution, but also impactsOSS/BSS vendors who have been evolvingtheir solutions to be more ‘NFV friendly’ inresponse to this, and opening up theirplatforms to be more modular, cloud-native and RestFul API driven.

It is critical to mention that apart from theproduction telco cloud (MANO and NFVI)environment, the telco should never missto build at least one of the followingMANO/NFVI environments: 1) atesting/prototyping environment, perhapsevolved from the lab built in the ExploringAge, and/or 2) a near-production stagingenvironment, to ensure and automate thereliable and predictable deployment ofVNFs and their updates.

Also, in this age, each domain (e.g. the vCPE solution and, optionally, itsunderlying classical network) will tend to have its own management andorchestration (MANO) platform and NFVI,and each one will communicate with theOSS/BSS systems individually, like newnetworks. The main goal of this ‘initially-silo-but-controlled-architecture’ is a fastertime-to-market for lucrative NFV/SDN usecases, and ramping up experience and

drive to transform the telco’s OSS/BSS tothe next-generation OSS/BSS. However, to limit and control the number of theseMANO/NFVI combinations, the telcoshould start adding more VNFs per MANOplatform in this age, which will also kick-start the multi-VNF cloud paradigm (trialing the DevOps approach), fosterinteroperability and add a level of healthycompetition.

Silo architecture for multiple domains

OSS/BSS

Management and orchestration (MANO)

NFV Infrastructure Virtualized Infrastructure Manager

Element Management

VNFs

Or-Vnfm

Vi-Vnfm

NF-Vi

Ve-Vnfm-vnf

Ve-Vnfm-em

Or-Vi

Med

iati

on

VNF Manager(s)

NFV orchestrator

NS Catalog VNF

Catalog NFV

Instances NFVI

resources

OS-

Ma-

nfvo

Branch site 1

Branch site 2

Enterprise Data Center vCPE

vCPE

WAN SDN Controller

Data Center

vCPE

MANO 3

vIMS

MANO 2

vEPC

MANO 1

NFVI

Regional DC

Internet

IP/MPLS NFVI NFVI

Figure 12: Silo architecture with integration between the MANO and the classicalinfrastructure per domain

Figure 13: Siloed development of SDN and NFV use cases using existing infrastructure (DC)

18


The aim and spirit of the Becoming Age istwo-fold: 1) Unleashing all the accumulatedexperiences, lessons learned andinternal/external buy-ins; and 2) Movingfrom ‘using digital technologies forextending existing capabilities’ to ‘usingthem to create new capabilities’.

Cloud computing is not a new concept butis gaining footholds in telcos, sped up byNFV and SDN technologies, and playing anessential role in the cloudification journey.By having a cloud infrastructure (NFVI)closer to customers, telcos will be able todeploy new sets of NFV and SDN scenariosthat enable new innovative services withspecial requirements (e.g. low latency andhigh throughput) that were complex orunfeasible before. However, in this age,Etisalat is learning that what can be moreimportant than location are twointertwined cloud computing tenets:elasticity and resource pooling. A cloud is awaste if its resources cannot be virtuallypooled, and it is not a cloud if it cannot beelastic within that pool. That is whyhomogeneity becomes key in theBecoming Age. In other words, the telcocloud graduates to becoming multi-tenantand agnostic to the VNFs. We hadrecommended earlier in the Doing Age tobegin shifting the telco cloud to beingmulti-VNF (even if as a silo/vertical), inpreparation for this. Now, the silos shouldstart integrating or defragmenting,preferably in a multi-vendor supplyarrangement. Strategically, it can also bemulti-VIM (i.e. multi-cloud managers). Withharmonising the NFVI, kick-starting theintegration or merger of the siloorchestrators, as well, is now due.

Telcos already host their IT applications,enterprise applications and supportsystems that are positioned in centralfacilities capable of providing infrastructurewith optimal cost, performance anduptime as per Tier (I-IV) specifications. Thenetwork or telco datacentres, on the otherhand, are distributed to host network ortelecom VNFs and SDN controllers thatprovide services closer to the customers.Hence, they should spring up in areas withpotential business and subscriber growth.So, unlike IT datacentres and IT clouds, adifferent strategy should be followed whenbuilding telco datacentres and telcoclouds.

In any case, the lessons learned by nowbecome confidently more specific. Forexample, in most of Etisalat’s NFVIdeployments, rack-mounted servers werefound to be more flexible than bladeservers—despite the opposite being truefor IT clouds. Another lesson was thatdeploying the VIM (e.g. OpenStack) in a

disaggregated, centralised architecture ismore suitable to distributed NFV than thehyper-converged, distributed architecture.

Moreover, by embarking on thecloudification journey, telcos can leveragelessons learned from other industries thathave been generating value from digitaltechnologies. When it comes to cloudcomputing, Amazon Web Services (Graph2) is by far the market leader, perhaps asbig as the next 10+ cloud providerscombined. Most analyses to understandthe root cause of Amazon’s predominancestrongly indicate that it is simply becausethey were early adopters. That is, theiraccumulated experiences, lessons learnedand internal/external buy-ins, whichAmazon had started accumulating earlierthan anybody, gave them the significantedge. We therefore expect the telco cloud,being a new and niche cloud, to similarlyreward the early adopters (i.e. telcos andvendors).

The becoming ageTransforming telcos using cloud computing

Niche players

Challengers

Graph 2: Gartner ’s IaaS magic quadrant representation released in June 2017

Low High

Abili

ty to

exe

cute

Lo

w

Hig

h

Completeness of vision

Visionaries

Leaders Amazon web service

Microsoft

Google

Other

players

Launched February 2010

Launched April 2008

Launched July 2002

19


The distributed infrastructure will enabletelcos to separate the control plane fromthe user plane in order to provide bettercustomer experience and unlock newservice possibilities as emphasised earlier.In this age, in addition to the nationaldatacentre (that should have been built inthe Doing Age), telcos should startdeploying regional datacentres as per themaster planning strategy, driven by solidbusiness cases. Based on this distributedNFVI, telcos can continue deploying NFVand SDN use cases as per their up-to-dateroadmap initiated in the Exploring Age(vCDN, vSBC, vDPI, vUDC, vPCRF, DC-SDN,DCI-SDN, vHLR, vFirewall). It now becomesimportant to address the rules to selectwhich hierarchy is better for eachdeployment. Some functions and usecases prefer to stay centralised, someothers require further distribution, whileothers fit into a multi-tier deployment ofnational, regional and edge DCs. Thechoice of a two-tiered, three-tiered or x-tiered distribution of DCs depends on thetelco’s strategy and its physical presence,but it has to be decided in the master

planning, and communicated to the entiretelco’s engineering and architecture cadre.The figure below illustrates an end-gameexample within a national geography(global or international DCs may beconsidered, or collapsed onto the nationalDCs). For instance, the nationaldatacentres might host telecom signallingfunctions, service orchestration and non-latency-sensitive functions, applicationsand systems that are storage and/orprocessing hungry, benefitting from highlycentralised pooling, like historicalanalytics/machine learning, mastercontrollers and digital service platforms.The national DCs need to be able to standby for one another. Note that even if onenational DC is planned, a DR site and planmust be included. Depending on themeasured latencies, the regional DC wouldcater to functions and applications thatcarry the data plane (customer-transmitteddata) that requires lower latency (e.g. lessthan 25ms), plus high-throughputfunctions (e.g. vPGW, vCDN, BNG). With theedge DCs, it is necessary to highlight thatlatency is not only related to the network

but also associated with the dataprocessing. Functions requiring extremelylow network latency (10ms or faster)and/or extremely low processing latency(for real-time actions) will need to be at theedge, avoiding the processing latencyincurred when transmitting all dataunnecessarily to central DCs andburdening them with hard-to-prioritiseprocessing. Cloud RAN, 5G, IoT, augmentedreality, V2X, healthcare, and safety-relatedapplications would be in dire need of theedge DCs. An additional side benefit ofedge DCs is the further distribution offunctions, enabling a more granular, on-demand and forecastable capacityplanning per edge DC, reducing softwareoverprovisioning buffers (e.g. licences) andincreasing the telco’s agility (e.g. granularexpansion or shrinking, with zero humantouch being possible).

Spreading the telco cloud towards the customer

CPE

RAN

CDN

OLT

MEC

APP

Edge DC

DC-SDN

Regional DC

MSC

P/S-GW

EPDG

CDN

BNG

SBC

CPE

APP

DC-SDN

SD-WAN

APP

NG-OSS/BSS

Service Orch.

MME

PCRF

HLR/HSS IMS

AAA

CDN

DNS

T-MSC

National DC

DC/DCI-SDN

T-SDN

Illustrative Illustrative Illustrative

Figure 14: The NFV and SDN functions and use cases hosted by distributed, hierarchical datacenters

20


With the spreading of the cloudification ofthe network, new concerns arise. A seriesof distributed datacentres are now hostingpieces of the “network intelligence” andtherefore need to be synchronised by anentity that ultimately will abstract thewhole underlay (e.g. optical transport) as asingle virtual network. The datacentresneed to be efficiently interconnected sothat balancing of service workloadsbetween facilities can be completelymanaged, optimised and eventuallyautomated.

The challenge is to make severaldatacentres spread across multiplegeographies to look like a single resourcepool, available for all network functionsand services. The DCI-SDN is a use casecapable of solving this issue in a scalable,agile and flexible way. To interconnect thedatacentres, a WAN composed of whiteboxes is of course required. Onetechnology available today to manage thetraffic among datacentres is based onvirtual extensible LAN (VXLAN), which is anoverlay data plane standard for datacentrenetworking. It can be viewed as a tunnelbetween two end points, which providesthe capacity for multi-tenant VPNs,improved resources allocation andprotection from topology or technologychanges [28]. Google’s use case of cloudDCI is probably the most known andsuccessful, with Google generating morethan 25% of all Internet traffic and themajority being handled by this newarchitecture. It also helped them pushsome network links to sustainably run atnear 100% utilization for long periods [29].It uses four components, developed byGoogle itself, built specifically for cloud

DCI: Espresso, which deals with peerconnections to ISPs; Jupiter, responsiblefor the traffic within datacentres; B4, which addresses connections betweendatacentres; and Andromeda, an NFV stackthat provides monitoring and managementcapabilities [30]. Another technologyenabling DCI-SDN uses transport layers,i.e. layers 0 (DWDM, photonics) and layer 1(SDH and OTN), which can also double asan SDN use case applicable to thetransport layers of the service providerWAN. Both technologies can complementeach other to enable DCI-SDN.

As the DCI-SDN plays a role at thedatacentres’ interconnection, the DCN-SDN is implemented to bring out thesoftware flexibility and economy frominside the datacentres onto each other by

decoupling the control plane from theunderlying network even further. Througha centralised management and trafficprocessing, the datacentre energy costscould be reduced by 30% to 40% [31]according to related studies and use cases.These savings are highly compelling in abusiness case since datacentres aresignificant consumers of energy.

A solid cloud DCI/DCN architecture is a must-have in this age for telcos that want to abstract their multi-vendorinfrastructure into a single cloudenvironment for end-to-end, multi-topology control, management andautomation. In addition to the agilitybenefits, costs efficiencies will ramp up topay off its implementation.

The becoming ageAbstracting the NFVI into a single cloudinfrastructure

DCI-Controller DCN-Controller DCN-Controller

Openstack NFVO Openstack

OSS/BSS

RestFul RestFul

RestFul RestFul

DC2

DC-GW

TOR

DC1

DC-GW

TOR

Netconf/OpenFlow

DCI-PE DCI-P DCI-PE DCI-P

DCI

Netconf/PCE/BGP-LS

VxLAN110 VxLAN210 VXLAN VXLAN L3VPN over RSVP TE

DWDM

RestFul

Netconf/OpenFlow

Figure 15: SDN solution to abstract the NFVI as a single cloud infrastructure

21


Another great area of interest for Etisalatis to see the possibility of centralization ofits operations and solutions by leveragingcloud technology. Yet again, telcos shoulddraw inspiration from Internet companiessuch as Google and Microsoft, who havebenefited greatly by adapting a centralisedoperation model for their services andcertain business functions.

Etisalat is of the opinion thattelecommunication groups who haveoperations in multiple countries mustexplore this idea to improve their businessoperations and competitive position in themarket. Etisalat terms this idea of unifiedproduction of services and certain networkservices “cloud factory”, which refers to anintegrated cloud platform and serviceswhere a centralised production of itsnetwork and operations support functionscan be harnessed with local networkentities in various operational companieswhich are situated in the Middle East,South Asia and Africa.

Traditionally, most of the operators grouphave focused primarily on thecentralization of operational services andprocurement processes. However, recentlysome operators have already made thecentralization of certain network andbusiness support functions a key part oftheir digital transformation strategy. Withcloud factory, Etisalat is not only hoping tooptimise its cost structure but also toembrace a competitive edge with theunique ability to launch innovative digitalservices at a much faster rate across itsmarket. This will also enrich synergy amongits operational companies, enabling thesharing of best practices and highly skilledresources. Having conducted a successful

proof of concept last year to prove thetechnical feasibility of cloud factory, Etisalatis now in the process of finalising itsstrategy related to it.

There are many obstacles which hinder thedeployment of this concept, most notableof which are the regulatory requirementsof “end-user data sovereignty”. There iscertainly a call from many industry playersto re-evaluate regulatory terms related tothe centralised deployment of networkfunctions for telecommunication. Other

challenges related to the technical maturityof integrated cloud-native based softwarefor certain network and operationalfunctions are also a concern for Etisalat.Lastly, Etisalat thinks global operatorsshould decide on a central operativemodel and make careful consideration ofall technical, organizational and regulatoryconcerns. If developed withcomprehensive planning, a business modelbased on the cloud factory concept isassured to be highly beneficial for digitaltransformation.

An integrated cloud for Etisalat International

• Agile / On-Demand Operations across OpCos

• Time to Market - (From Months to Days)

• Time to Expansion – (From Months to Hours)

• Unified Service Experience across OpCos

• Unified Version Management

• Unified Brand Value

• Hardware/Software Resources Sharing among OpCos

• DC Consolidation with Distributed Virtual DC Setup

• OPEX Optimization with unified efficient operations

• Unified Cloud Platform for VAS

• Unified Cloud Platform for Digital Services

• Platform for Public Cloud Offering

Time to market

Unified service experience

Cost optimization

Revenue growth

Cloud factory group level

OpCo A

OpCo B

OpCo C

Figure 16: Cloud factory architecture and benefits for a group level perspective

22


The ultimate target for telcos is to evolvetowards a more service-oriented, efficient,and agile operating model. Therefore,transforming all VNFs and the telco cloudto be ‘cloud-native’ is seen as one of themost important steps (and even a finalfrontier) in the cloudification journey.Specifically, this refers to telecomcapabilities to bundle together smallpieces of software (aka “micro-services”) tocreate services on-demand with minimumresource obligations, and as per end-userrequirements. This form of services isusually referred to as “stateless” and it ispossible to deploy these servicesanywhere in the cloud with minimumhuman touch. The modular service willgrow or shrink as required, and continuesto consume or relinquish resourcesefficiently as per its need. This requires ascalable, elastic, and programmablenetwork cloud with closed loopautomation of all operational and businesssupport systems.

A growing number of industry players arealready pursuing this cloud-nativeparadigm and it is gaining momentumacross all standards bodies, open-sourceorganizations and consortiums. Worthhighlighting, two technologies havebecome the poster children of cloud-nativeapplications, although they may conflictdown the line: containers (e.g. Kubernetes),and the more advanced serverlesscomputing (e.g. the proprietary AmazonLambda).

In any case, telcos in this age are movingaway from vertical integrations andembracing common, agnostic platformswhich will ease the journey to stateless

services and unlocking the true potential ofcloud computing. However, telcos haveanother justifiable concern in that telecomVNFs’ behaviours and requirements arevery different from IT workloads. Hence, asimplistic approach of juxtaposing the ITexperience onto a telecom environmentwould not cut it in this context.

In what we call the ‘first wave ofvirtualization’, VNFs are a little more than’ported’ or ‘re-packaged’ versions of theirclassical/physical ancestors, withhandicapped cloud abilities. It may well bemore challenging for many incumbentvendors to re-write their success storiesfrom scratch to develop stateless-basedsoftware than to virtualise them in the firstplace. Nevertheless, the ‘second wave ofvirtualization’ has started to produce morecloud-based VNFs that avail more of what

the telco has to offer. Fortunately, some ofthe incumbent vendors are committed tobecoming cloud-native in their strategyand roadmaps as a necessity for thedevelopment of their future products.Although standardised specifications of‘cloud nativity’ still lack clarity amongstmost industry players, differentapproaches are being passionatelypursued towards non-monolithic softwaredevelopment. This will trigger a ‘third waveof virtualization’ in which VNFs are writtenfrom the base to be truly cloud-native.Etisalat believes that it is vital for telcos inthis age to aggressively increase theirdemands for cloud-native VNFs andapplications, reducing their tolerance ofported or repackaged VNFs occupying theresources and capabilities of their telcocloud.

Embracing cloud native

VNFs

… VNF 1 VNF 2

Automatic deploym

ent

Infrastructure agnostic

I f

Telco DevOps platform

VNFs

On-demand scaling & deployment

Assembling micro service module

Close loop automation

NFV

NFC (Cloud native)

Figure 17: The transition to a cloud native environment

23


Scalability and elasticity are the hallmarksof cloud computing. A couple of NFV andSDN use cases can help with that.Virtualization resources will be needed inthe edges of the network, covering a widergeographic area and ultimately enablingtelcos to deploy new services to theircustomers. Thus, more efficient utilizationof these resources will be required. Thesescaling use cases include the vRAN (radioaccess network), vOLT, vBGN and the T-SDN (transport controller). In order todeploy these technologies, the NFVI againneeds to be ready, and the edgedatacentre layer might make sense to beleveraged to host these virtualisedfunctions and local SDN controllers. Bylogically centralising the control andmanagement of the whole backbone andbackhaul into the T-SDN controller, telcos

will be able to provision resources in anagile and automated way, dynamicallyallocating and releasing network resources(backhaul/backbone/metro) whennecessary. The vRAN can generateoperational and cost efficiencies by poolingBBUs that will manage and control mobileresources through the RRUs, alsosignificantly increasing the spectrumutilization, which is an expensive andscarce asset for telcos. By virtualising thefunction of vOLT, flexibility is also unlockedat the FAN (fixed access network).Scalability is also a key factor for both usecases, since the BBU and OLT functionsare virtualised based on software. Thearchitecture combining all these conceptsis illustrated below:

Scaling up the virtualization technologies

National DC

Regional DC Regional DC Mobile access

Fixed access

T-SDN

WAN SDN

RRU

RRU RRU

Edge DC

vRAN

ONT

ONT

Edge DC

vOLT

WAN SDN

By reaching the edge of the networkinfrastructure with thedistributed telco cloud, it will be possible toadopt edge computing or multi-access edgecomputing (MEC),pushing the intelligenceand processing closer toend-customers to unlockinnovative servicepossibilities.

Figure 18: NFVI architecture reaching the edge of the network (fixed and mobile)

24


Multi-access edge computing (MEC), fogcomputing, or otherwise known simply as‘edge computing’, has emerged as one ofthe most significant technology trends. Itpaves the way for harnessing the fullpotential of future services like V2X, VR/AR,tactile internet and others, as pointed outpreviously. Additionally, it is also beingpromoted by some analysts as having anintegral part in 5G deployment scenarios.Edge computing as a concept refers todeploying certain real-time processing,content distribution and intelligencefunctions in close proximity to the end-users, pushing intelligence and processingcapabilities down where the dataoriginates and is consumed. This not only

helps in meeting the requirements for the aforementioned services, but it alsoreduces any unfavourable investmentnecessities and traffic engineeringcomplexities towards the network core.With the availability of a distributed NFVI,the MEC technology can be executed atthe edge layer of the cloud infrastructurein an environment characterised by ultra-low latency and high bandwidth.

MEC or fog computing technologies can beadopted in the Being Age since the telcowill have a distributed NFVI by now, wherethe processing can be done at the edgelayer of the telco cloud infrastructure.

Edge computing appliedat the newtelecom NFVI

National cloud

Regional cloud

Edge cloud

SOC-UP/SOC-CPSOC-UP/SOC-CP

SOC-UPSOC-UP

FMC-GW MCDN

IoT -Core Exposure agent

AR/VR

MCDN

E2E MPSlice Mgt.

Resource Mgt.

X-GW 3rd APP X-GW

D-GW D-GW Exposure agent

MEC (Multi-accessedge computing)

IoT MSA / FMC Wi-Fi Fixed Wireless

2G/3G/4G/5G Stadium Enterprise Public Sec.

Diverse Applications

SOC-CP / SOC-EP

Policy Mgt.

UDB Mobility Mgt.

Slice Orchestration

Session Mgt.

SOC-UP

Figure 19: MEC application in a distributed infrastructure architecture (NFVI)

25


The MEC technology mandates seriousefforts towards new specifications andstandards. ETSI’s MEC framework is a goodstart, but it has gaps to close beforecommercial adoption. Etisalat is of theopinion that the whole ICT industry shouldwork more closely and swiftly towardsstandardization of MEC’s detailedspecifications. Key considerations of MECdeployment that still need to be addressedin standards are the integration with thecore network, billing and authentication,enabling lawful interception, how toincorporate 5G architecture (access-agnostic) and how to conduct trafficsteering. Moreover, our view is that MECshould be considered as a ‘solutioncontainer’ rather than a solution itself. That

is to say that it should have the capabilityto embrace several solutions, functions,APIs and applications from differentvendors/third parties offering smoothintegration. In parallel to the MECstandardization, the specifications from3GPP are evolving; for instance, the currentrelease 3GPP R14 related to CUPS (control-user plane separation) does not supportMEC requirements—a topic that will beaddressed at the next release 3GPP R15.Etisalat has an expectation to have at mostin 2018 a standardised MEC technologyready to be launched. Thisaccomplishment will only be possible withboth standardization bodies (ETSI and3GPP) working together and actively withthe industry.

Edge computing: specifications andchallenges

MEC ISG launch

07/2015

12/2014

2015Q3

CUPS, SI 2016Q3

CUPS, SA2, freeze 2016Q2 CUPS, WI

2018Q2 QR15 WI SA3 freeze

CCF, slice, programming

2019Q4 R15 SA3 freeze

3GPP R15

3GPP R14 CUPS 2017.6

CUPS stage 3 freeze

2016.2

3GPP R16

2015Q4 QR15 SI

CCF, slice, programming

2018 R16 SI

2017Q12016Q2 2018Q2

R15/R16 is for 5G

Phase 1 Standardization Phase2 Standardization

Figure 20: MEC standardization Vs 3GPP standardization timeline

26


Imagine ordering a book through thewebsite of a digital bookstore. The onlineorder will set off a cascading chain ofautomated service orchestration functions.No human from the digital bookstore willget on the phone to call the deliverycompany for the delivery of the book.Everything is automated. Unfortunately,many business-to-business (B2B) networkservices are still manually provisioned andrequire human intervention. This is evenmore so with telecom services. You cannotorder an MPLS circuit like you order abook. The process of ordering the circuit isstill long, manual and painful, and in direneed of automation. As a result, newservices may take days, weeks or evenmonths to be fulfilled.

Achieving zero-touch provisioning meansautomating the entire value chain so thatwhen a customer wants a service, they justhave to press a button on the company’swebsite to begin the process ofprovisioning, and after a few minutes theprocess is complete. The service can thenbe used and even adjusted on demand,rather than requiring a locked-in servicecontract. This concept has been in theworks by entities such as MEF through itsspecifications for the lifecycle serviceorchestration (LSO) approach [8], and TMForum, which has a project named Zero-touch Orchestration, Operations andManagement (ZOOM) that is focused onthis topic [32]. According to TM Forum, toachieve zero-touch provisioning, operatorsmust first close the loops. This meanscollecting and analysing data to figure outhow the network can be optimised, andthen implementing those changes in anautomated way [33]. Later, telcos mayadopt the LSO, which is basically an end-

to-end agile approach to simplify andautomate the service lifecycle in asustainable fashion for coordinatedmanagement and control across allnetwork domains responsible fordelivering a full service.

By using a digital marketplace representedin the figure below as a self-service app orportal, customers will be able to requestand customise the services as per theirneeds, and telcos will provide for suchrequests automatically through the serviceorchestrator. To achieve such a vision,telcos should guarantee that networkorchestration is done in real-time andwithout human intervention. Independentof the type of orchestrator, it must becapable of simplifying the user experiencevia self-ordering for a complex service, withzero-touch provisioning and full integrationwith the critical business and operationprocess across a hybrid network with bothphysical and virtualised (NFV)

infrastructure, including virtualisedcustomer premise equipment (vCPE).

By closing the loop and implementing theservice orchestrator, specialised staff willbe free to focus on complex activitiesrequiring their expertise, instead ofmonitoring the infrastructure, doingrepetitive tasks and taking routinedecisions. This will allow telcos to not only reduce OpEx and react faster tounplanned changes in the infrastructure,but also to focus more on innovation andoptimizations. However, the specificationsand practical use cases from the serviceorchestrator are far from being matureand ready to deploy.

The service orchestrator enabling zero-touchprovisioning

Partner


Business applications

Service orchestration

Telecom


Business applications

Service orchestration

Self-service web portal

Headquarter Cloud service provider

Figure 21: The goal to provide E2E management and control across multiple network domains

27


Service orchestration arose mainlybecause today’s OSS/BSS have not beendesigned and built for future digitalproducts and service demands. They werebuilt for static and predictable telecomservices instead (figure 19).

Hence, a new generation of OSS/BSS (NG-OSS/BSS) could be developed to deal withthe new elements, such as the MANO, SDNcontrollers and also the serviceorchestrator, to facilitate the agile designand creation of new services and enablereal-time, data-driven and on-demandprovisioning and billing. If properly done, itwill reduce time to design and on-boardnew services to just days instead of theaverage of 10 months in some telcos [35][36].

Furthermore, the new systems should also be API-driven (in order to guarantee a seamless integration with the serviceorchestrator) and must have the ability to provide visibility across all networkdomains and vendors so they can supportfulfillment and assurance of services thatare constantly changing [37].

Therefore, the existing structure andsystems need to be transformed to realisethe full benefits of this age. The evolutiontowards a zero-touch provisioningenvironment requires a new systemarchitecture to guarantee the integrationof NG-OSS/BSS with ‘self-service portal’,‘orchestration’ and ‘marketplace’components. These key blocks will nowown and be responsible for typical OSS/BSS functions that will be re-arranged inorder to better connectcustomers/partners with services/productsand resources/network management withthe ecosystem orchestration and policies(figure 20).

Integration of service orchestrator withNG-OSS/BSS

Strategy,infrastructureand product

Operationalsupport and

readiness

Fulfillment Assurance Billing

Customer

Service

Resource

Product

Market / sales

Supplier & partner

Enterprize domain

1

2

3

4

5

6

Customer/ partner orchestrationlayer

Service orchestration layer

Resource orchestration layer

OSS

Resource management

Network management

Converged chargingand billing

Product catalog and customer/partner mgt.

VNFsapplicationsmarketplace

Controllers layer

Network layer

Intelligent policymanagementE2E service management

and orchestration

NFV orchestration

Policy rules prediction analytics

monitoring closed control

loop

Self-service portal B2B gateway

Customer Partner

Service management resource

1 3 4 2 6

26323

4

5

5 5

4

Figure 23: To-be systems and orchestration organization (based on TMF’s ZOOM framework)

Figure 22: As-is support system organization (based on TMF’s TAM framework)

28


In order to transform to being a digitalservice provider, it is necessary to buildintelligent systems that are self adaptableby harnessing the enormous datagenerated to enhance the customerexperience. NFV and SDN have enabledthe digitalization of the telecomenvironment and transformed it fromhardware-centric to software-centric,which has led to exposure to huge oilfieldsof data. In order to drive meaning andextract the context and value out of the bigdata, it is necessary to build digital,intelligent systems that learn fromexperience and make real-time decisionswhenever required to address customerrequirements efficiently, throughsupervised or even unsupervised machinelearning. This is different to traditionalautomation where humans hard-code thepolicies or bots replicate what a humanwould do. Machine learning is just that—itsprimary purpose is to learn whileexecution of the learnings can bedelegated to appropriate systems like theservice orchestrator and digital platforms.Moreover, the ever-growing use of mobileand data being moved to the cloud willforcefully demand the integration of NFVand SDN with big data and ArtificialIntelligence (AI) in this age.

Additionally, ETSI has created a workinggroup named Experiential NetworkIntelligence (ENI) to develop a cognitivenetwork management architecture for theoptimization of the telcos operationalenvironment. Furthermore, incorporatingAI technology in today’s systems that cananalyse millions of data can help makepredictions to provide unique services toeach customer (mass customization).Therefore, an efficient AI-powered big dataand analytics ecosystem can help telcosimprove quality of experience by analysing

network traffic in real time, conductingpredictive capacity planning, identifyingfraudulent behaviours and anomaliesinstantly by analysing call data records,personalising marketing campaigns toindividual customers using real-timelocation and social networks data, anddeveloping new services unique to eachcustomer. This can result in higherbusiness velocity and increasedprofitability for telcos, which is not feasiblevia manual ways. [36]

In a telco cloud paradigm, where all theprocesses and the network itself aredynamic, the efficient use of the availabledata related to customers in every serviceand product decision will be a key successfactor. A telco that incorporates suchmethods will gain a real competitiveadvantage and generate increasedrevenue.

The fusion of AI with NFV and SDNtechnologies to build digital intelligentsystems will enable highly optimised andefficient resource utilization and costefficiencies for telcos and customers alike.In addition to that, it will self operationalisethe systems that require little or no humanintervention, and can heal, scale and adapton their own, taking the customerexperience (and digital telco) to the nextlevel.

Integration of NFV and SDN with Big Dataand AI

Data ingestion and pre-processing

Data processing service

Data storage Computing Data assets

management

Advanced analytics

Knowledge insights

Real-time decision

Services development

Information analytics

Data openness services

Catalog User admin. Billing and charging

Data open API

Infrastructure

Eco system enablement

Business valuecreation

Operational efficiency

We believe telcos will notbe able to hire staff fastenough to do thingsmanually to cater to thenew suite of digitalservices. That is whyanalytics and AI are keyfor us.

Figure 24: Telecom big data analytics and its benefits generated

29


30


“Cloud computing presents uniqueopportunities for Etisalat to attainsustainable growth in the emerging digitalecosystem. Our aim is to deliver adifferentiated service experience toempower a digital and happier society. Thisnew opportunity-rich landscape calls forcollaboration with diverse industry playersfocusing on open and agile operating andorganizational models. Those who have yetto start this journey, must start now andrespond boldly with the right strategy!”

Considerations from Etisalat’s Sponsors

“Customer needs, behaviours and evenperspectives are changing constantlyacross all demographics and generations—now more rapidly than ever. Web start-upsand Internet corporations became good atseizing these opportunities. And telcoshave to evolve faster to catch up. That iswhy we are cloudifying our network—tomake it agile and adaptive to the maximumwhile maintaining profitability. How elsecould you be a 5G, an IoT or a digitalprovider?“

Hatem BamatrafEtisalat InternationalChief Technology Officer

Esmaeel AlhammadiEtisalat UAESenior Vice President NetworkDevelopment

31


Pedro TavaresDeloitte PortugalTelecom Engineering Centre ofExcellence (CoE) Leader

“At Deloitte we believe that digitaltransformation is not only abouttechnology: deep transformations mighthappen in telcos at multiple dimensions inorder to revamp their business. We alsobelieve that a “big bang” approach has alower probability of succeeding; instead,the journey might be travelled step by stepand harvest benefits along the way thatwould, in turn, help to fund more complextransformations. That being said, thispaper focuses on the technologicalperspective, since it represents thefoundation to support all otherdimensions. Additionally, in terms ofinvestments and effort, it might be theheaviest expenditure for telcos in thedigital transformation.

During the development of this whitepaper, we were able to corroborate ourhypothesis that telcos are just scratchingthe tip of the iceberg with virtualizationand cloud technologies, not having yetreaped all of the promised benefits. Thissituation might lead to the completediscredit of these exponentialtechnologies. Clearly, the main point thathas to have the pace stepped up regardsthe standards, which are not mature

enough. The Cloudification journey is partof telcos’ digital transformation and it is theroad to reach the envisioned end-statescenario. It has such a high importancethat telcos and their partners (vendors,consultancies) must be more proactivewhen it comes to investing and allocatingefforts with the standardization bodies toevolve the technologies that compose thepathway. There is an evident gap ofengagement (very high to none) fromdifferent stakeholders in thetelecommunication industry.

Deloitte will invest effort to push forwardthe frontiers of knowledge related tovirtualization technologies (SDN/NFV) untilthey reach a viable maturity level for ourclients. We will support our clients toaccomplish their vision, providing agnosticadvisory in terms of the best suitablesolutions for their businesses in additionto address the challenges across all layersfrom the operating model.”

Our contributors’ point of viewConsiderations for Etisalat

32


Peng XiongjiPresident Etisalat Key AccountHuawei Technologies

“There is no doubt that the emergence of a digital lifestyle and services is re-shaping the industry landscape and‘transformation’ has become inevitable for service providers. The potential of‘cloudification’ and SDN/NFV technologieshas certainly raised the expectations of theindustry with a practical answer to theimpending challenges of increasingservices deployment time, cost and rigidand inefficient operational model. It isencouraging to see Etisalat taking thestrategic step to embark on its digitaltransformation journey.

Network transformation should beconsidered as an opportunity to deploy an“experience oriented” agile network andoperating model rather than an imperativecompulsion to address short-termbusiness challenges.

Huawei deems that it is of extremesignificance to have an unambiguous andcomprehensive plan upfront beforeinitiating a network transformationprogram, especially one related to thedeployment of distributed NFVinfrastructure and virtualised networkservices with operational automation.

Telcos should be aware of the distinctionbetween hyped expectations and realisticgoal-setting. Therefore, it is of criticalimportance to form trustworthy long-termpartnerships with solution providers whohave an in-depth familiarity with theintricacies of both the IT andcommunication industries. While doing so,carriers should always keep an “enricheddigital experience” as the top-most driverand fundamental criterion for any futurenetwork design and re-architecture.

Huawei, as an active contributor to theindustry advancement, and long-termstrategic partner of Etisalat, is fullycommitted to assist Etisalat on thisdigitalization path and would feelprivileged to work with Etisalat to helpthem achieve their goals.”

33


34


About usAuthors and contributors

Khaled IsmaeelAbdullaEtisalat InternationalVice President FixedNetworks

Authors

Contributors

Ali (AlKaff) AlHashmiEtisalat UAEEmerging Technologiesand Centre ofVirtualisationExcellence

Sajeel WaqasSenior ConsultantTelco TransformationHuawei Technologies

Guilherme OliveiraManager Deloitte M.E.Engineering Centre ofExcellence (CoE)

Liu JingHuawei TechnologiesSolution Architect

Ranjan SinhaTechnology Strategy &Architecture Services(TS&A)

Tiago PiresConsultant DeloittePortugalEngineering Centre ofExcellence (CoE)

Dr. GuoleiHuawei TechnologiesSDN/NFV Chief Expert

Shahid Imran KhanHuawei TechnologiesNetwork SolutionManager

André SantiagoSenior ConsultantDeloitte South EastAsia EngineeringCentre of Excellence(CoE)

Syed Mohi u DinHuawei TechnologiesNetwork Consultant

Pedro TavaresPartner DeloittePortugalLeader EngineeringCentre of Excellence(CoE)

Esmaeel AlhammadiEtisalat UAESenior Vice PresidentNetwork Development

Hatem BamatrafEtisalat InternationalChief TechnologyOfficer

Emmanuel DurouPartner Deloitte M.E.Leader TechnologyMedia andTelecommunications(TMT)

35


Further detailsIndex of acronyms

AAA Authentication, authorizationand accounting server

AEB Automatic emergency breakAPI Application programming

interfaceAR Augmented realityB2B Business-to-businessB2C Business-to-customerBBU Baseband unitBNG Broadband network gatewayBSS Business support systemsCAGR Compound annual growth rateCDN Content delivery networkCEM Customer experience

managementCGNAT Carrier grade NATCOTS Commercial-off-the-shelfCPE Customer premises equipmentCRM Customer relationship

managementCSP Communications service

providerDC DatacentreDCI-SDN Datacentre interconnect SDNDHCP Dynamic host configuration

protocolDNS Domain name serverDPDK Data plane development kitDPI Deep packet inspectionDSP Digital service providerDWDM Dense wavelength division

multiplexingE2E End-to-endEBIT Earnings before interest &

taxesePDG Evolved packet data gatewayFAN Fixed access networkFMC Fixed-mobile convergenceGPON Gigabit passive optical network

GW GatewayHLR Home location registerHSS Home subscriber serverHW HardwareIaaS Infrastructure-as-a-serviceIMS IP multimedia subsystemIoT Internet of thingsISP Internet service providerLAN Local area networkLB Load balancerLSO Lifecycle service orchestrationMANO Management and orchestrationMCDN Mobile CDNMEC Multi-access edge computingMGW Media gatewayMME Mobile management entityMSC Mobile switching centreNAT Network address translationNG Next-generationNFV Network function cloudificationNPV Net present value NFV Network functions virtualizationNFVI NFV infrastructureNFVO NFV orchestratorOCP Open compute projectOLT Optical line terminalOSS Operations support systemsOTT Over-the-topPCRF Policy and charging rulesfunctionPNF Physical network functionPoC Proof-of-conceptPSTN Public switched telephonenetworkQoE Quality of experienceRAN Radio access networkRRU Remote radio unitSaaS Software-as-a-serviceSBC Session border controller

SDN Software-defined networking SDP Service delivery platformSD-WAN Software-defined WANSOC Security operations centreSW SoftwareTCO Total cost of ownershipT-SDN Transport SDNUDC User data convergenceUI User interfaceVAS Value-added servicesvCPE-R Virtual CPE residentialvE-CPE Virtual enterprise CPEVIM Virtual infrastructure managerVNF Virtual network functionVNFM VNF managerVoLTE Voice over LTEVoWiFi Voice over WiFiVPN Virtual private networkVR Virtual realityV2X Vehicle-to-everythingVXLAN Virtual extended LANWAN Wire area network

36


References

[1] I. Godlovitch et al., “Over-the-topplayers (OTTs)”, Directorate-Generalfor Internal Policies – EuropeanParlmament, 2015

[2] “The Mobile Economy – Europe2015”, GSM Association, 2015

[3] “The Mobile Economy – Africa 2016”,GSM Association, 2016

[4] “The Mobile Economy – NorthAmerica 2016”, GSM Association,2016

[5] “The Mobile Economy – LatinAmerica and the Caribbean 2016”,GSM Association, 2016

[6] “The Mobile Economy – Middle Eastand North Africa”, GSM Association,2016

[7] “The Mobile Economy – Asia Pacific2016”, GSM Association, 2016

[8] P. Lee et al., “Technology, Media andTelecommunications Predictions2017”, Deloitte, March 2017

[9] “Digital Maturity Model & Metrics”,TMForum, May 2017

[10] “OpenStack launches asIndependent Foundation, BeginsWork Protecting, Empowering andPromoting OpenStack”, BusinessWire, September 2012

[11] “Network Functions Virtualization(NFV); Use Cases”, ETSI GS NFV,October 2013

[12] “ZOOM Project”, TMForum,“https://www.tmforum.org/collaboration/zoom-project/” (Last visited inJune 2017)

[13] “Lifecycle Service Orchestration(LSO): Reference Architecture andFramework”, MEF, March 2016

[14] M. Rouse, “White box server”,TechTarget, January 2016,“http://searchdatacentre.techtarget.com/definition/white-box-server”(Last visited in May 2017)

[15] “Data Plane Development Kit”, Intel,“http://dpdk.org/” (Last visited inJune 2017)

[16] Open Compute Project,“http://www.opencompute.org/about/” (Last visited June 2017)

[17] Gerben Steunebrink, “TheServitization of product–orientedcompanies”, University of Twente,BiZZdesign, August 2012

[18] D. Bushaus, “NFV: Bridging TheChasm”, TM Forum, February 2016

[19] G. Ragoonanan, G. Yigit, “vCPEservices business case: potentiallybillions of dollars payback for fixedCSPs”, Analysis Mason, October2015

[20] “AT&T Universal Customer PremisesEquipment (uCPE)”, AT&T, June 2016

[21] I. Fukuda, P. Roychowdhury,“Delivering a carrier-class NFV use-case”, NTT Group, OpenContrail,May 2015

[22] Telefonica and NEC win innovationaward for virtualised home gatewaytrials”, GSMA, January 2015,“https://www.gsma.com/membership/telefonica-and-nec-win-innovation-award-for-virtualised-home-gateway-trials/” (Last visited inJune 2017)

[23] S. Buckley, “Verizon to offer virtualCPE services in 30 markets by theend of 2016”, FierceTelecom,“http://www.fiercetelecom.com/telecom/verizon-to-offer-virtual-cpe-services-30-markets-by-end-2016”(Last visited in June 2017)

[24] A. Rao, G. Ragoonanan, “The pivotalrole of service assurance systems inNFV/SDN environments”, AnalysisMason, June 2015

[25] “vEPC for Communications ServiceProviders”, Intel, April 2016

[26] “Etisalat successfully launches itsfirst virtual Telecom function, acloud-based mobile core”, AffirmedNetworks,“http://www.prnewswire.com/news-releases/etisalat-successfully-launches-its-first-virtual-telecom-function-a-cloud-based-mobile-core-300301282.html” (Last visited inJune 2017)

[27] S. Atreyam, “Economic Benefits ofvirtualised Evolved Packet Core”,IDC, Cisco Systems, June 2016

[28] D. Noguer Bau, “SD-WAN, NFV’s bestFriend”, J-Net Community,“https://forums.juniper.net/t5/SDN-and-NFV-Era/SD-WAN-NFV-s-best-friend/ba-p/298359” (Last visited inMay 2017)

[29] J. Skorupa, A. Lerner, C. Canales,“Forecast: SD-WAN and Its Impacton Traditional Router and MPLSServices Revenue, Worldwide, 2016-2020”, Gartner, November 2016

37


[30] P. Scott, “Four Reasons Why SD-WANMakes Sense”, sdnwanresource,“http://www.sdwanresource.com/articles/419405-four-reasons-why-sd-wan-makes-sense.htm” (Last visitedin May 2017)

[31] J. Burke, “The CIO’s Guide to SD-WAN”, Nemertes Research, April2017

[32] M. Ardica, P. Bellagamba, Y. Louis,“VXLAN Multipod Design for IntraData Center and GeographicallyDispersed Data Center Sites WhitePaper”, Cisco, May 2017

[33] S. Jain et al., “B4: Experience with aGlobally-Deployed Software DefinedWAN”, Google Inc, 2013

[34] L. Hardesty, “Google brings SDN tothe Public Internet”, SDxCentral,April 2017,https://www.sdxcentral.com/articles/news/google-brings-sdn-public-internet/2017/04/ (Last visited: June2017)

[35] ZOOM team, “Business Rationaleand Technical Overview forOrchestration and AutonomicControl Loops”, TM Forum,September 2016

[36] D. Bushaus, “NFV Catalysts: Solvingreal-world problems,”,tmforuminform,“https://inform.tmforum.org/features-and-analysis/2014/11/nfv-catalysts-help-service-providers-solve-real-world-problems/” (Lastvisited in May 2017)

[37] “OSS/BSS could be the bottleneckthat limits SDN and NFV benefits”,Global Telcos Business, May 2014,https://www.globaltelecomsbusiness.com/article/b11vydvlty3tpv/ossbss-could-be-the-bottleneck-that-limits-sdn-and-nfv-benefits (Last visited inJune 2017)

[38] F. Ruhl, S. Steiff, S. Latawa, P. Moore,J. Hawkins, T. Cauble, G. Feeney,“Impact of SDN and NFV onOSS/BSS”, Open NetworkingFoundation, March 2016

[39] B. Fox, R. van den Dam, R. Shockley,“Analytics: Real-world use of big datain telecommunications”, IBMInstitute for Business Value, SaïdBusiness School at the University ofOxford, 2013

[40] “Future OSS”, Orange & Huaweiwhite paper, February 2017

38


Index of figures and graphics

Figure 1. Annual variation (CAGR) forecastfor mobile networks from 2015 to 2020

Figure 2. Main drivers for Etisalat’scloudification journey

Figure 3. Use cases for value-addedservices with special network requirements

Figure 4. Order of magnitude of human reaction times

Figure 5. Digital Tranformation Framework (Based on TMFORUM)

Figure 6. Ilustration of The digitaltransformation journey and its challenges

Figure 7. The cloudification journey’s ages

Figure 8. High-level envisionedinfrastructure and system scenario

Figure 9. NFVI high level architecture

Figure 10.Worldwide standardizationbodies and foundations landscape

Figure 11. Lab environment with SDN andNFV combines architecture

Figure 12. Silo architecture withintegration between MANO and theclassical infrastructure per domain

Figure 13. Siloed Deployment of SDN andNFV use cases using existing infrastructure(DC)

Figure 14. The NFV & SDN functions anduse cases hosted by distrivuted,hierarchical datacenters

Figure 15. SDN solution to abstract theNFVI as a single cloud infrastructure

Figure 16. Cloud Factory architecture and benefits for a Group Level perspective

Figure 17. The transition to a Cloud NativeEnvironment

Figure 18. NFVI architecture reaching theedge of the network (fixed and mobile)

Figure 19. MEC application in a distributedinfrastrusture architecture (NFVI)

Figure 20. MEC standardization Vs 3GPPstandardization timeline

Figure 21. The goal to provide E2Emanagement and control across multiplenetwork domains

Figure 22. As-is support systemsorganization (based on TMF’s TAMframework)

Figure 23. To-be systems andorchestration organization (based on TMF’sZOOM framework)

Figure 24. Telecom big data analytics andits benefits generated

Graphic 1. NFV/SDN Functions MaturityAssessment

Graphic 2. Gartner’s IaaS Magic Quadrant representation released in June 2017

39


This publication has been written in general terms and therefore cannot be relied on to cover specific situations;application of the principles set out will depend upon the particular circumstances involved and we recommend thatyou obtain professional advice before acting or refraining from acting on any of the contents of this publication.Deloitte & Touche (M.E.) would be pleased to advise readers on how to apply the principles set out in this publication totheir specific circumstances. Deloitte & Touche (M.E.) accepts no duty of care or liability for any loss occasioned to anyperson acting or refraining from action as a result of any material in this publication.

Deloitte refers to one or more of Deloitte Touche Tohmatsu Limited, a UK private company limited by guarantee(“DTTL”), its network of member firms, and their related entities. DTTL and each of its member firms and their relatedentities are legally separate and independent entities. DTTL (also referred to as “Deloitte Global”) does not provideservices to clients. Please see www.deloitte.com/about to learn more about our global network of member firms.

Deloitte provides audit, consulting, financial advisory, risk advisory, tax and related services to public and private clientsspanning multiple industries. Deloitte serves four out of five Fortune Global 500® companies through a globallyconnected network of member firms in more than 150 countries and territories bringing world-class capabilities,insights, and high-quality service to address clients’ most complex business challenges. To learn more about howDeloitte’s approximately 245,000 professionals make an impact that matters, please connect with us on Facebook,LinkedIn, or Twitter.

Deloitte & Touche (M.E.) is a member firm of Deloitte Touche Tohmatsu Limited (DTTL) and is a leading professionalservices firm established in the Middle East region with uninterrupted presence since 1926. DTME’s presence in theMiddle East region is established through its affiliated independent legal entities which are licensed to operate and toprovide services under the applicable laws and regulations of the relevant country. DTME’s affiliates and related entitiescannot oblige each other and/or DTME, and when providing services, each affiliate and related entity engages directlyand independently with its own clients and shall only be liable only for its own acts or omissions and not those of anyother affiliate.

Deloitte provides audit, tax, consulting, financial advisory and risk advisory services through 25 offices in 14 countrieswith more than 3,300 partners, directors and staff. It is a Tier 1 Tax advisor in the GCC region since 2010 (according tothe International Tax Review World Tax Rankings). It has also received numerous awards in the last few years whichinclude best Advisory and Consultancy Firm of the Year 2016 in the CFO Middle East awards, best employer in theMiddle East, the Middle East Training & Development Excellence Award by the Institute of Chartered Accountants inEngland and Wales (ICAEW), as well as the best CSR integrated organization.

© 2017 Deloitte & Touche (M.E.). All rights reserved.

Documents

From pipelines to clouds Etisalat’s playbook€¦ · services require special capabilities that today’s network infrastructure cannot provide. This is where telco’s emerging