3. NFV and SDN Guide
for Carriers and
Service Providers
Technology Overview, Benefits,
and Target Applications
Essentials
4. 4
Contents
Executive Summary.......................................................................................................................................... 5
Roadmap for this Book.................................................................................................................................... 6
NFV and SDN Fundamentals....................................................................................................................... 6
NFV Basics.......................................................................................................................................... 6
What Is NFV?..................................................................................................................................... 7
How Does NFV Work?.................................................................................................................. 8
Why Is NFV Needed?.................................................................................................................... 8
Why NFV Now?................................................................................................................................. 9
The NFV Ecosystem...................................................................................................................... 9
SDN Basics...................................................................................................................................... 11
What Is SDN?.................................................................................................................................. 11
SDN Tenets...................................................................................................................................... 12
Why Is SDN Needed?................................................................................................................. 13
Building Blocks of a Virtualized Network............................................................................................. 14
Orchestration Layer.................................................................................................................... 15
Control Layer.................................................................................................................................. 15
Infrastructure Layer.................................................................................................................... 16
NFV and SDN Building Blocks Summary........................................................................ 16
The Value Proposition for NFV and SDN............................................................................................ 16
Benefits of NFV and SDN........................................................................................................ 17
Summary.......................................................................................................................................... 19
Open Source Projects and Standards for NFV and SDN ......................................................... 20
SDN and NFV Use Cases............................................................................................................................ 21
Use Case 1: Enterprise-to-Cloud Migrations............................................................... 21
Use Case 2: Lowering Operational and Capital Costs............................................ 22
Use Case 3: Service Providers and Transform Their Businesses .................... 24
Use Cases Summary................................................................................................................. 24
What’s Ahead for SDN and NFV.............................................................................................................. 24
Glossary................................................................................................................................................................ 26
Acronyms............................................................................................................................................................. 27
5. 5
Executive Summary
Service providers’ sense of urgency to use Software-Defined Networking (SDN) and
Network Functions Virtualization (NFV) is being driven by the demand for dynamic
services and customer control. The drive to virtualize the network is different from
other transformational initiatives in the past, as the impetus for change is based
on end-user demands rather than technological advancements by networking
vendors. Service providers are always looking for ways to cut Operating Expenses
(OPEX) and Capital Expenses (CAPEX) while delivering networking solutions that
support their ever-changing business goals. A perfect storm of new technological
capabilities—increasing demands imposed on networks due to cloud computing
and big data, growing market acceptance for network virtualization, and the avail-
ability of NFV and SDN products—sets the stage for an explosion in demand for
SDN and NFV by service providers around the world.
NFV and SDN represent a revolutionary way of configuring, deploying, and main-
taining networks that provides a compelling business case for service providers to
fundamentally change the way they implement customer-facing network services,
functions, and capabilities. SDN decouples the data plane from the management
plane, enabling network orchestration, control, and management in a way that is not
possible using traditional networking hardware and software solutions. NFV, mean-
while, promises to virtualize network services, decoupling software from proprietary
hardware. This allows network operators to quickly configure and deploy new net-
work services using a policy-based management paradigm running on commodity
hardware. Both technologies support highly agile networking, elastic provisioning,
reduced time-to-revenue for new services, and frictionless customer turn-ups.
Collectively, NFV and SDN present service providers with an opportunity to imple-
ment a more customer-centric network infrastructure, where the network easi-
ly—and, in some cases, automatically—adapts to dynamic customer needs and
requirements. By delivering programmability, network agility, and simplified network
management based on open concepts and standards, NFV and SDN help custom-
ers transition to software-defined networks based on commodity hardware and
open software to lower costs and avoid vendor lock-in.
6. 6
Roadmap for this Book
This NFV and SDN essentials guide will give readers a concise overview of the archi-
tectural concepts of NFV and SDN. It will also discuss market drivers and the value
propositions for NFV and SDN, including how NFV and SDN can increase a compa-
ny’s revenue while reducing costs. This guide will outline the foundational building
blocks of NFV and SDN, and, lastly, discuss the future of NFV and SDN from Ciena’s
perspective. This document also includes a glossary of general networking terms
and acronyms for reference.
NFV and SDN Fundamentals
NFV Basics
As IT infrastructure, including compute and storage functions, continues an inevita-
ble march towards virtualization, networking is the next logical area where network
operators can virtualize to keep pace with technological change. As applications in-
creasingly tax networks for bandwidth, flexibility, and speed, the notion of overbuild-
ing networks to accommodate peak traffic loads becomes untenable and unafford-
able. It is no longer acceptable to purchase application-specific hardware, engineer
and configure it for that unique application, and expect it to be in-service for 10
years or more. What is needed is greater agility and control of the network and its
core functions. This virtualization implies programmability using software control.
Following the recommendations of the European Telecommunications Standards
Institute (ETSI) and the NFV Industry Standards Group (ISG), NFV has emerged as
the means to virtualize network functions, ideally over an SDN.
Today, many network functions are implemented as special-purpose, custom-built
devices. These devices have custom hardware, firmware, and chipsets that help
accelerate performance. With NFV and SDN, more and more of that same function-
ality is being implemented in software rather than in hardware. Thanks to Moore’s
Law, which focuses on the concept of diminishing returns, network functions that
previously were possible only via highly customized hardware and software can now
be implemented totally in software. This basic fact drastically changes the network
landscape for service providers and other network operators.
NFV gives service providers and operators the opportunity to lower their network
infrastructure costs while speeding up the configuration and deployment of new
network services. This new, more flexible, software-based network service envi-
ronment allows service providers and operators to quickly spin up new network
7. 7
services as needed for specific situations and customers, shortening the process
from the weeks or months to days or even minutes. This business agility creates a
significant competitive advantage because it allows network operators to pursue
new markets and opportunities that were not economically viable using traditional
networking hardware and software, and do so much more quickly.
What Is NFV?
NFV can reduce or eliminate application-specific, proprietary hardware from the
network infrastructure. Instead of requiring a network operator to deploy new
hardware components for each new service or application, the NFV model provides
the same functionality using virtual appliances running on commodity x86 servers.
Conceptually, an operator can deploy a virtual appliance on demand and upgrade its
functionality through software updates over time. Figure 1 presents a comparison of
the classic network model versus the new network model.
Figure 1. Classic network model features hardware-based appliances and components, while
the new virtual network model utilizes virtual appliances and components. (Image source:
ETSI) Traditional physical networks incorporate specialized hardware that offers one physical
node per role, manual installation processes per site, and static, hard-to-scale operations.
Meanwhile, virtualized networks are software based. Multiple roles may run on the same hardware
components. Virtual networks enable remote operation and management. They are dynamic and
easy to scale, based on operational requirements.
8. 8
Firewalls, Provider-Edge (PE) routers, Deep Packet Inspection (DPI), and encryption
are a few examples where currently an operator typically deploys separate hardware,
often from different vendors. Each hardware element faces its own obsolescence
cycle, requires its own certification activities, and needs to be managed, often in
unique ways. Managing these multiple vendors and functions is complicated and
fundamentally too expensive to scale and support. Worse, the ability to quickly turn
up new applications is difficult to impossible for most service providers.
How Does NFV Work?
In the NFV model, virtual appliances residing on physical or virtual servers replace
dedicated hardware-based network appliances. Operation and administration
functions are handled through an orchestration system that coordinates the virtual
appliances in operation on a network. Like virtual machines, virtual appliances are
selected based on end-customer needs and deployed as needed when and where
they are required. Scaling to adapt to changes in customer need is based on loading
software onto the appropriate servers. Further, when the virtual appliance is no
longer needed, the space on the server can be freed up for use by another applica-
tion. The sharing of resources in this way helps Service Providers to drive down their
overall costs.
Why Is NFV Needed?
NFV promises to simplify the physical network architecture while improving its ability
to scale and adapt to technological change. For example, a regional bank currently
supports its branch locations through an IP-Ethernet network that demands edge
routing, encryption, and Ethernet switching between its corporate campus and
regional or local branch locations. Currently, the bank would purchase, qualify, and
manage three separate hardware elements to achieve this functionality. Costs
include CAPEX to purchase each item, OPEX for qualification testing and installa-
tion, and then ongoing OPEX to manage the services. Each element also consumes
valuable physical space and energy in a data center or wiring closet. Leveraging
NFV, the bank’s network service provider would be able to provide managed routing,
encryption for data in motion, and virtualized security services to the bank at a very
attractive price point, with superior capabilities. A single commodity server could be
installed at the bank’s corporate campus, and the application software for each of
the three applications would be downloaded over the service provider network, and
execute on the single server. There is even the option for the network service pro-
vider to provide some applications as managed services that execute on a server at
the network operations center, instead of at the bank’s premises.
9. 9
Further, as functionality evolves or standards change—for example, in encryption or
virtualized security—updates are made through software remotely, without physical-
ly touching any hardware.
As service providers and enterprise customers alike continue to demand better
offerings at lower costs, NFV introduces the concept of highly agile, software-based
network functions that increase responsiveness to end-user requirements while
being easier to manage and faster to deploy. NFV offers distinct competitive ad-
vantages over traditional network services. NFV can also give end-users greater
control over their own Wide Area Network (WAN) services, including enabling elastic
provisioning of network functions, increasing network agility, enhancing security,
and providing considerable cost savings compared to dedicated appliances.
Why NFV Now?
NFV promises to not only improve the scalability and agility of a service provider’s
operations, but to do so while reducing networking costs. In the example above,
where the network operator is providing virtualized services to a regional bank,
capital expenses are reduced through fewer, cheaper, and less frequent hardware
purchases. Operational expenses are also reduced through lower requirements for
physical space and energy and, to a lesser extent, through shorter vendor qualifi-
cation and interoperability testing requirements. In place of hardware capital costs,
the bank pays for the software-based virtual functions they selected for their branch
locations. As the needs of their business change due to changes in customer hab-
its, locations of bank branches, or the addition of new banking services, the bank’s
virtual network can change as needed by deploying exactly the network services
they require.
The NFV Ecosystem
With NFV, network operators can approach the solution to their IT needs much like a
trip to the grocery store—buying what is needed, when it is needed, from a selec-
tion of different vendors. Service providers can purchase and quickly deploy only
the amount of network resources required based on customer demands. Gone are
the days when a service provider had to stock a large amount of single-purpose
network equipment in anticipation of increasing customer demand that may or
may not materialize. Less risk and expense translates into more revenue and profit
for the service provider. Once the network operator has deployed a generic hard-
ware appliance at the customer’s premises, the customer can shop for the Virtual
Network Functions (VNFs) that will create the virtual appliances needed for network
operations.
10. 10
The VNF ecosystem includes a wide variety of components from multiple vendors.
A few examples of VNFs include vRouter functionality, security and encryption,
load-balancing, virtual Set-Top Boxes (vSTB), WAN optimization, and performance
monitoring. Once selected, the VNFs are controlled and operated through what
ETSI has defined as the Management and Orchestration (MANO) function. MANO
includes the distribution of VNFs across hosts, orchestration of VNF functionality,
and management of a VNF’s lifecycle. Figure 2 shows a selection of vendors offering
VNFs and virtual appliances.
NFV MANO consists of three functional areas that accomplish all tasks related
to the lifecycle of a VNF: NFV Orchestrator (NFVO), VNF Manager (VNFM), and
Virtualized Infrastructure Manager (VIM). The NFV orchestrator brings on new
network services and VNFs and provides global resource management, validation,
and authorization of VNF infrastructure resource requests. VNFM controls specific
VNF instances, coordinating infrastructure resource requests between the VNF
instance and related network element management systems. The VIM controls
and manages NFV infrastructure, which includes compute, storage, and network
resources.
Operators have multiple options in terms of the servers used for hosting VNFs,
NFVO, and VIM, and their physical locations. Depending on service provider options
and enterprise strategies, servers could be virtual resources located in a cloud data
center, a central office, a dedicated pod, or the customer’s premises.
Figure 2. A selection of vendors that develop and market VNFs
11. 11
SDN Basics
SDN has rightfully garnered much recognition over the last few years as the tech-
nology has rapidly progressed from conceptual architecture to standards-based
products and capabilities. SDN is a disruptive technology that continues to mature
as new products are introduced into the market. And the momentum behind SDN
continues to grow as users create new and innovative applications that combine
SDN with NFV.
What Is SDN?
SDN is a network architecture based on the concept of separating management
and control of a network from its physical hardware. This logical separation enables
network behavior to be programmed by a diverse set of applications and services
using open APIs. By opening up traditionally closed networks, SDN allows service
providers to control networks and devices in a consistent manner, regardless of the
underlying hardware or networking technology (for example, Carrier Ethernet (CE),
IP/MPLS, or SONET/SDH). By driving standards for the programmatic interfaces,
SDN applies unified management and control to these formerly disparate networks.
SDN is now being recognized as a logical solution to a variety of scale, security,
management, and administration issues that have proved to be challenging with
networks using conventional networking technology. Service providers and enter-
prise customers see benefits to the flexibility and programmability offered by SDN,
Figure 3. A high-level view of the SDN architectural layers and communication
paths, indicated by the black arrows (Source: ONF)
12. 12
but in reality, end-user demand is also pushing companies to adopt SDN. Mobile
computing, Bring-Your-Own-Device (BYOD), and hybrid cloud architectures can
drastically shift network traffic characteristics. Conventional networking archi-
tectures cannot easily adapt to these ever-changing requirements, while SDN is
specifically designed to handle these types of dynamic, demanding network envi-
ronments.
SDN Tenets
SDN improves network control and reduces the cost of a network while increasing
agility by leveraging four key attributes:
1) Network programmability – SDN enables network behavior to be controlled
by software residing outside the networking devices that provide physical
connectivity. As a result, network operators can tailor the behavior of their
networks to support new services and even individual customers. Program-
mability enables a new wave of automation to dramatically streamline oper-
ations. By decoupling the hardware from the software, operators can rapidly
introduce innovative, differentiated new services free of the constraints
imposed by closed and proprietary platforms.
2) Logically centralized intelligence and control – SDN architectures benefit
from the availability of logically centralized network topology and state, to
enable intelligent control and management of network resources. Tradi-
tionally, network control methods are distributed, with devices functioning
autonomously, providing operators with limited awareness of the state of
the network as a whole. By centralizing control, elements such as bandwidth
management, restoration, security, and policy can be made highly intelligent
and optimized, exploiting a domain view of the network. This also leads to a
high level of service quality and hence high customer experience.
3) Abstraction of the network – Services and applications running in an SDN
network are abstracted from the underlying technologies and hardware that
provide physical connectivity and network control. Applications will interact
with the network through APIs instead of management interfaces tightly
coupled to the hardware.
4) Openness – SDN architectures usher in a new era of openness, enabling
multi-vendor interoperability and leveraging tools from a vendor-neutral eco-
system independent of any single vendor. Openness is provided from above,
13. 13
where open APIs support a wide range of applications, including cloud
orchestration, OSS/BSS, SaaS, and business-critical networked apps; from
below, where intelligent software controls hardware from multiple vendors
through open programmatic interfaces; and from within, where intelligent
network services and applications can run in a common SDN software envi-
ronment.
Why Is SDN Needed?
Essentially, all service providers intend to deploy SDN. In fact, many have moved
beyond test phases and are beginning to deploy the technology in their production
networks. The benefits of SDN are wide-ranging, but operators have three main
priorities for migrating their network infrastructure to SDN:
1) Drive new revenues through operational agility. Operational agility enables
new services to be introduced faster and Service Providers can be more
responsive to customer requirements
2) Reduce OPEX through by streamlining operations and incorporating auto-
mation.
3) Reduce CAPEX, not necessarily through inexpensive hardware, but rather by
optimizing the usage of resources like bandwidth and data center network
capacity as well as by reducing space and lowering power consumption.
A key advantage of SDN is the ability for network operators to build applications and
programmatically control their networks though SDN-enabled APIs. SDN enables
applications that are not just network-aware, but can intelligently monitor network
conditions and automatically adapt the network configuration as needed. Thus, SDN
enables agility for the delivery of dynamic and on-demand services like bandwidth
on demand. SDN also facilitates network automation, which reduces operational
costs for deployment, management, and operations. The cost savings realized
through SDN-based automation both increase profits and reduce costs while
achieving the ultimate benefit of improved agility.
Another benefit of SDN is the ability to establish policies for managing networks,
service-levels, and applications, rather than explicitly managing network configu-
rations for physical resources. Network policy management is made possible in an
SDN environment because the SDN controller collects global knowledge of the net-
work. That global knowledge allows SDN controllers to map policy settings across
14. 14
the network and into the appropriate configuration of each network device. This
capability enables SDN operators to leverage policies to manage networks from a
services perspective—such as users and applications—as compared to the phys-
ical resource-based approach used by legacy network architectures. Policy-based
automation frees up network administrators from mundane activities so they can
concentrate on more important activities like troubleshooting complex network
issues or network planning.
SDN automation also has the potential to help service providers find and retain
qualified network administrators. By automating operations, network complexity is
reduced and network managers can manage a greater number of devices with few-
er errors and greater productivity. Changes take far less time, and resources may
be utilized more efficiently than in the past. In the IT space, server administrators
have leveraged virtualization and automation to the extent that the typical server
administrator can effectively manage multiple thousands of servers. SDN will give
network administrators that same sort of leverage, to manage many hundreds or
thousands of network devices, rather than dozens. SDN exponentially expands the
scope and reach of network operations personnel, reducing CAPEX and OPEX in
the process.
Building Blocks of a Virtualized Network
Identifying and understanding the components of a virtualized network and how
they interrelate is key to understanding the path to a successful SDN and NFV im-
plementation. Virtualized networks include an Orchestration Layer, a Control Layer,
and an Infrastructure Layer. Each of these components provides crucial capabilities,
interfaces, and features that enable transformation to a programmable, open, and
automated network.
• Orchestration Layer – multiple domains, automation
• Control Layer – control of individual domains
• Infrastructure Layer – physical or virtual networks and resources
These building blocks combine to create a solid foundation for virtualized networks
to offer robust, automated, agile services to service providers, operators, and enter-
prises. Figure 4 shows the three layers of the virtualized network model: the Orches-
tration Layer, the Control Layer, and the Infrastructure Layer.
15. 15
Orchestration Layer
With the transition to virtualized networks, the Orchestration Layer provides the crit-
ical function of abstracting services from the underlying infrastructure components,
including physical and virtual networks. The abstraction, combined with open APIs
enable service providers to create, deploy and manage services across multiple
network domains, and automate their lifecycles – a process which we refer to as
‘multi domain service orchestration’. In addition to interfacing with the Control and
Infrastructure Layers, the Orchestration Layer also leverages open interfaces and
APIs to communicate with traditional OSS and BSS systems to integrate with back-
end service management processes, such as service assurance, alarm monitoring,
order management, and billing.
The decoupling of network services from the actual underlying network and their
elements through a common abstraction layer reduces the need for custom coding
across different network and management siloes. With the right service orches-
tration solution, Service Providers benefit from their ability to add, configure and
provision new devices and services flexibly and in realtime, to meet the require-
ments of today’s on-demand services -- even in the most complex, heterogeneous
environments.
Control Layer
The Control Layer is focused on controlling and managing individual physical and
virtual domains, providing the programmability and intelligence to manage and
control traffic flows within that environment. The Control Layer includes platforms
such as underlay and overlay SDN controllers, cloud or VIM platforms, and NFV
orchestrators that control and configure network resources. The Control Layer also
Figure 4.
16. 16
features standards-based interfaces to the north and south that tie the Orchestra-
tion Layer to underlying network functions, and open interfaces to the east and west
that support scaling of the virtual network.
Key requirements for the Control Layer include an approach and framework that
leverage open standards to ensure interoperability with other hardware and soft-
ware. Another key requirement for the Control Layer is the support of legacy
network protocols such as CLI, TL1, and SNMP, in addition to newer protocols and
interfaces such as REST APIs, OpenFlow, and Netconf/YANG.
Infrastructure Layer
The Infrastructure Layer includes the physical and virtual appliances and devices
that comprise the network layer in a virtualized network. The Infrastructure Layer
is where the processing and forwarding of data occurs. In a multi-domain environ-
ment, these network components can include both virtual networking and legacy
network gear that has been integrated into the Infrastructure Layer. The Infrastruc-
ture Layer also provides intelligent forwarding, switching, and routing protocols that
can be reconfigured at any time without requiring the network to be brought down
for maintenance. As network intelligence moves from network devices such as
proprietary routers and switches to the Control Layer, network hardware becomes
a commodity component, driving down hardware costs, while SDN automation
increases the agility of configuring and maintaining the network. The Infrastructure
Layer communicates with the Control Layer using legacy network protocols such
as CLI, TL1, and SNMP in addition to newer protocols and interfaces such as REST
APIs, OpenFlow, and Netconf/YANG.
NFV and SDN Building Blocks Summary
NFV provides service providers and enterprises with the advantages of increased
network agility and flexibility while offering lower costs to procure, deploy, and
manage innovative network services. The software-defined network is the founda-
tion upon which these services are delivered more efficiently and cost-effectively.
Providing open standards and industry consensus-based interfaces for these com-
ponents is crucial for the interoperability and efficient operation of networks using
SDN and NFV.
The Value Proposition for NFV and SDN
To foster market acceptance of NFV and SDN, the value proposition must be clear
and compelling to operators, service providers, vendors, and customers. For exam-
ple, compared to traditional network technology, NFV and SDN offer three distinct,
17. 17
well-defined competitive advantages for service providers: automation, optimiza-
tion, and monetization, as illustrated in Figure 5.
Automation improves operational efficiency while reducing the time to market for
new network services. Optimization conserves network resources while improving
the quality of network services. Monetization leverages intelligent pricing models
and the ability for mass customization to generate new revenue for operators. This
book examines the competitive advantages service providers can expect to derive
from NFV and SDN as crucial contributors to their ongoing success.
Benefits of NFV and SDN
NFV and SDN increase network agility and service velocity, which improves cus-
tomer satisfaction. It is not unusual for customer requests for new network services
to take weeks or even months for operators to implement. Obviously, long imple-
mentation delays due to the complexity and hands-on nature of network service
provisioning is not a palatable option for most customers. As long as all operators
are using proprietary networking gear, customers do not have many options when
they need faster deployment of new services and applications. But now that SDN
offers programmability and automation, that increased agility gives customers the
option of taking their business to operators who can deploy new services or make
service changes much faster by leveraging the capabilities of NFV and SDN. A
recent market survey by IHS Markit indicates that more than 80 percent of service
providers are interested in SDN because of the ability to deploy network services
much more quickly, the highest motivator listed in the survey, as highlighted in
Figure 6.
Figure 5. Automation, optimization, and monetization are key competitive advantages of SDN.
18. 18
The agility offered by SDN and NFV is a huge competitive advantage that can
increase revenue by delivering to customers what the competition cannot, faster
than ever before. Increased agility alone makes a strong case for SDN adoption
among operators. Fortunately for operators and customers, SDN has even more
benefits.
Network optimization: SDN allows operators to optimize their networks via
programming and custom applications. For example, as new networks are deployed
in an SDN infrastructure, software automatically computes the optimal paths
through the network. Since those optimal paths can degrade over time due to
subsequent changes in the network topology or configuration, the application can
also run periodically to automatically defragment the network and re-compute
optimal paths.
Figure 6. An IHS Markit survey, 2016 Carrier SDN Strategies for the SDN “Drivers” highlights what
service providers are looking for in NFV and SDN.
19. 19
Operational efficiencies: NFV and SDN promote operational efficiencies by auto-
mating routine network maintenance tasks, network provisioning, and troubleshoot-
ing. Automating network optimization, as described above, is an excellent example
of how operational efficiencies inherent in SDN can streamline routine network
management.
Refocusing on growth via SDN and NFV automation: As SDN and NFV work to
streamline operations, software will take over more of the day-to-day management
of SDN networks. As a result, existing network personnel can spend less time mon-
itoring and maintaining services and more time creating services and generating
revenue. This shift in utilization improves the efficiency of the business and allows
for budgetary reallocation to generate more profit.
Openness of SDN platforms: Clearly, operators have long bristled under the
restrictive environment offered by proprietary network vendors, being forced to
wait months or years for coveted features and capabilities to finally make it into a
released product. Proprietary hardware also tends to be expensive, and once an
operator commits to one networking vendor, it gets harder to consider other man-
ufacturers’ equipment due to the real possibility of compatibility and interoperability
issues. SDN assists with interoperability issues by using a robust set of standards
that migrates the majority of network intelligence from a dedicated, closed-architec-
ture network device to standard hardware and software available from a multitude of
OEMs.
Summary
The value proposition for SDN and NFV combines the competitive advantages of
programmability, the technical advantages of centralized intelligence, the finan-
cial benefits of automation, and the increased agility that helps service providers
and operators to attract new customers. SDN and NFV represent a profound shift
in the focus and control of networking technology, from proprietary vendors to
standards-based OEMs, creating hardware that works seamlessly with hardware
from other OEMs. A cursory inspection of SDN and NFV features makes it clear that
these specifications have been developed with operators, rather than networking
vendors, in mind. This stark difference from earlier network standards efforts is
exactly why SDN and NFV are not considered just an evolutionary step, but rather
a revolutionary leap forward for service providers. NFV plus SDN puts operators,
service providers, and enterprises in the network driver’s seat, with features and
capabilities designed to make operators more agile, competitive, and profitable.
20. 20
Open Source Projects and Standards for NFV and SDN
The specific paths to market for SDN and NFV have been molded and guided by
standards bodies and industry consortia. The SDN specification is being developed
by the Open Networking Foundation (ONF) standards organization, while NFV is
being formed and molded by the European Telecommunications Standards Institute
NFV Industry Specification Group (ETSI NFV ISG). Both SDN and NFV specifications
are considered to be frameworks for networking vendors who want their devices
and software to interoperate with other standards-compliant vendors, while indi-
vidual vendors are free to deviate from the specs as needed to differentiate their
products, offer capabilities beyond the specs, or ensure compatibility with legacy
network gear.
The ETSI NFV ISG includes networking software vendors, service providers, network
operators, national technology organizations from numerous countries, academic
and research institutions, and related user groups. The ETSI NFV ISG management
philosophy is based on open standards, and uses a consensus model for devel-
opment and definition decisions related to NFV. In this sense, the ETSI NFV ISG is
more of an industry cooperation model than a traditional standards organization
that might tend to be less transparent about the negotiations and compromises
that form the standard. Many NFV vendors and customers are currently leveraging
widely-used NFV APIs that are not part of the NFV specification, yet are critical to
the success of NFV. In this sense, standards are definitely important, but the goal is
a working NFV network architecture that goes beyond the functions and capabilities
of the “official” NFV spec. The ETSI NFV ISG recently announced a collaboration
with the Metro Ethernet Networking (MEF) standards body to integrate NFV as the
network services architecture for Carrier Ethernet 2.0, a developing standard for
carrier-scale Ethernet.
What has evolved into the SDN specification is actually a combination of work by
industry consortia and standards bodies such as the Open Networking Founda-
tion (ONF). Though the ONF has been instrumental in organizing the effort behind
formation of the SDN specification, parallel work by various industry groups has also
added momentum and vital market input to the network virtualization movement.
Evidence of collaboration of the many industry bodies includes the “An Industry Ini-
tiative for Third Generation Network and Services” whitepaper published in Novem-
ber 2016, authored by MEF, ON.Lab, ONOS, OPEN-O, OpenDaylight, ONF, OPNFV,
and TM Forum. This paper describes a vision for the transformation of network
21. 21
connectivity services and the networks used to deliver them, based on network-as-
a-service (NaaS) principles which make the network appear to the user as the user’s
own virtual network with bump-in-the-wire value-add services. The end goal is a
Third Network that combines the on-demand agility and ubiquity of the Internet with
the performance and security assurances of today’s business grade networks.
SDN and NFV are changing the role of standards organizations and industry groups
in open source projects and specifications. Compared to past standardization
efforts that seem to emphasize the standard itself, SDN and NFV are much more
focused on implementation and operational capabilities and agility, not just the
underlying paper standards. In other words, the specifications for SDN and NFV are
based much more on practical considerations and less on the theoretical applica-
tions that might be part of a standards organization’s definition efforts.
SDN and NFV Use Cases
Real deployment scenarios and numerous network studies have quantified the ben-
efits of deploying SDN for cloud migrations, cost reduction, and enablement for new
markets. The customer use cases below highlight the opportunities and benefits
offered by SDN:
Use Case 1: Enterprise-to-Cloud Migrations
SDN and NFV technology offers numerous capabilities and cost benefits for en-
terprises looking to move all or part of their IT infrastructure to the cloud. However,
that migration can be a challenge for enterprises moving from a world in which their
IT personnel completely control the enterprise environment, LAN, and computing
within their environment to a cloud-based approach, where some amount of control
over infrastructure is ceded to the cloud vendor. Yet the economics of cloud com-
puting and infrastructure compels enterprises to migrate suitable applications and
other IT functions to the cloud for budgetary reasons. Cloud computing is almost
always cheaper than large private enterprise implementations because much of the
underlying cloud cost structure benefits from economies of scale, and costs are
spread across multiple users of those cloud resources. While gaining cost efficien-
cies in the cloud, enterprises must also come to grips with the fact that the enter-
prise will not have control over network configuration and performance in the cloud.
In this use case, a hypothetical enterprise needs to connect multiple locations
across the WAN. Obviously, traditional WAN networking can accomplish that goal
today, but a new cloud-based architecture also must ensure cloud connectivity to
and from each of the sites. This pushes connectivity to the cloud from the local data
22. 22
center network; this requires comprehensive security, as the data now traverses
outside the corporate LAN. Proprietary, purpose-built hardware appliances sup-
port this new architecture today, but utilizing SDN and NFV is a much more flexible,
cost-effective solution in this use case.
To mitigate these cloud migration challenges, a network built on SDN and NFV can
address each of these concerns in a programmable, flexible, high-performance
corporate network. Using NFV, network functions such as firewalls, virtual routing to
and from the cloud, NAT, antivirus, Distributed Denial-of-Service (DDOS) protection,
and WAN optimization can all be virtualized. Similarly, SDN capabilities can be used
to orchestrate overall network operations and cloud connectivity across the sites.
SDN and NFV can also combine to provide the requisite security capabilities needed
to keep Personally Identifiable Information (PII) and other mission-critical corporate
data safe and secure.
In this use case, Ciena’s Blue Planet provides three critical capabilities to enterprises
migrating to the cloud using SDN and NFV:
1. SDN orchestration of networking across the WAN and to/from the cloud
2. Managing and controlling VNFs such as DDOS protection, firewalls, DNS,
NAT, and virtual routing across compatible virtual network devices from a
variety of vendors
3. The software and support that tie together this cloud-enabled network with
multiple company data centers and remote locations
Use Case 2: Lowering Operational and Capital Costs
The pressure to do things better, faster, and cheaper than the competition is a fact
of life for service providers. As a result, many operators are exploring the possibility
that transitioning to an SDN- and NFV-based network will lower their operational and
capital costs. Using NFV to virtualize network functions is cheaper and more scal-
able than buying more dedicated appliances and other purpose-built hardware. Now
that other options are emerging, operators are looking to SDN and NFV for a more
efficient means of building and managing their networks.
With networks built on dedicated hardware, operators must maintain a sizable fleet
of expensive trucks and personnel with deep technical expertise to keep their
networks running smoothly. The overhead for operators using traditional network
23. 23
hardware and architectures is immense. SDN provides a programmable, open inter-
face that can orchestrate software-defined devices, as well as legacy network gear
in most cases. Rather than physically installing new gear, SDN allows for download-
ing and provisioning of network services as needed. This ability to automate and
remotely administer far-flung physical networks reduces the need for hands-on
maintenance of network components, thus reducing the need for truck rolls.
Utilizing SDN, network technicians no longer have to physically inspect network
components or observe LED lights and aural alarms. Those health checks and
troubleshooting can usually be performed remotely with SDN. Similarly, with NFV,
operators no longer need proprietary hardware to provide network functions such
as DDOS protection, routing, firewall, anti-virus, and similar capabilities. Less physi-
cal hardware naturally leads to less time spent troubleshooting and maintaining the
network via manual inspection. The software-based capabilities of SDN and NFV
allow operators to reduce the size of those truck fleets and reduce the number of
skilled personnel dedicated to keeping the network operating. That lowers the oper-
ator’s underlying cost structure, which allows that operator to lower prices for their
customers, making them more competitive in the markets they serve.
In addition to automation capabilities, SDN and NFV also lower costs through better
end-to-end monitoring and better software tools for visualization and trouble-
shooting. These operational efficiencies allow operators to pursue customers in
market segments that were previously too small or unprofitable to make it worth
their while. Lower operational and management costs benefit the operator’s bottom
line, increasing profit potential and cash flow. As a result of lower costs, operators
that previously concentrated on serving larger businesses might now be able to
compete effectively for revenue from smaller businesses. Prior to the advancement
of SDN and NFV, these lower-tier customers were likely ignored by operators due
to the thin margins in that market segment. Lowering network costs now makes
those lower-margin customers a viable market for operators looking to expand their
customer base.
The case for cost efficiencies due to the use of SDN and NFV provides two distinct
benefits for operators:
1) Reducing OPEX and CAPEX costs provides increased profitability.
2) Reducing cost structures opens up new markets by turning previously un-
profitable customers into an opportunity for increasing market share.
24. 24
Use Case 3: Service Providers and Transform Their Businesses
The third use case highlights how large-scale connectivity companies, enterprises,
and service providers can leverage SDN and NFV to transform their business mod-
els by lowering costs and pursuing new customers and market opportunities. One
high-performance connectivity provider has traditionally competed in the enter-
prise and service provider data center interconnect market. This provider installed
an SDN-enabled WAN and secured direct connections to various cloud partners
such as Amazon Web Services, Windows Azure, and Google Cloud. By lowering
costs and streamlining the process for provisioning and managing their WAN cloud
interconnects, this provider can now resell their cloud-direct access to enterprises
that could not otherwise afford or justify a direct connection to a cloud provider. By
offering these enterprises the opportunity to enjoy the performance advantages of
being direct-connected to their cloud(s) of choice, this provider has opened up vast
new markets for its connectivity services.
Use Cases Summary
The world of data center computing in recent years has been changing far more
rapidly than that to which conventional networking can possibly adapt. SDN was de-
veloped to address the demands posed by the proliferation of mobile devices, Bring
Your Own Device (BYOD), cloud computing, server and storage virtualization, and
the increasing demands on networks to be flexible and dynamic. Service providers
can leverage SDN to drive down costs while increasing customization capabilities
and customer satisfaction, leading to CAPEX and OPEX savings. Enterprises are
also eager to realize efficiencies of scale and reduction of OPEX and CAPEX to
make their companies more competitive. Early SDN adopters will have a competitive
advantage over their conventional counterparts while developing the expertise to
leverage SDN even further as the standards evolve. As SDN matures, operators are
facing a convergence of virtualized servers, storage, networks, and applications that
will eventually be managed by a common set of tools and techniques.
What’s Ahead for SDN and NFV
Now that the underlying foundation of SDN and NFV is solid and networking vendors
are shipping commercially available SDN and NFV products, the next few years will
likely show a marked increase in the adoption of these technologies. Most service
providers already have SDN and NFV technologies either in the proof-of-concept
stage or in production. As companies, network architects, and network engineers
gain more experience with these technologies and become comfortable with the
new way of doing network things, all signs indicate that adoption will continue ex-
panding from service providers into the enterprise IT space.
25. 25
Nothing breeds success like success, so the more companies that successfully
make the transition to SDN and NFV, the more their competition will be motivated to
consider adopting SDN and NFV as well. Thanks to the compelling business case
for SDN and NFV, competition will likely be a strong motivator for the widespread
adoption of SDN and NFV. The pro-adoption argument includes greater agility and
control of the network, opportunities for revenue uplift, streamlining of network
operations, and differentiation of service offerings. SDN and NFV are disruptive
technologies in the network market, as evidenced by the large number of network-
ing startups and networking OEMs rethinking how to transform their product lines to
thrive in a virtualized network world.
That said, some challenges remain for SDN and NFV, including the inevitable
learning curve required to master this new approach to networking. There are also
preliminary discussions about integrating SDN into optical domains instead of only
supporting packet domains, as it currently does. Also, as SDN and NFV mature as
technologies, the end-to-end orchestration of these new technologies—including
integration with legacy networking technologies during the early phases of most
SDN implementations—will continue to be a challenge. Recognizing that orches-
tration is the glue that binds together SDN and NFV in large-scale networks, Ciena
is preparing a follow-up book entitled The Ciena Essentials Guide to Orchestration.
This will give readers an overview of orchestration, from design to implementation
and maintenance, as they continue their SDN and NFV educational journey with
Ciena and Blue Planet.
26. 26
Glossary
Controller
A logically centralized component of software providing network management (and
network control) functionality. For example, a controller establishes the policies
and rules regarding packet forwarding and configures the network infrastructure to
perform the forwarding.
Domain
A networking component with centralized management. A domain can be as small
as a single network function or device or as broad as an entire network.
OpenFlow
A communications protocol for programmatically controlling the forwarding plane
of a network switch or router. OpenFlow assumes a separation of the control plane
from the data plane and directs packet flow through pattern-match/action com-
mands.
Orchestrator
A software component used to provide end-to-end control across a network. An
orchestrator is generally considered to provide higher-level perspective than a con-
troller. For example, it may provide end-to-end service abstraction, where a control-
ler might focus on packet forwarding and control.
Pod
A self-contained unit that includes compute and storage. NFV pod refers to an inde-
pendent x86-based component dedicated to running VNFs.
Python
A widely used, high-level, general-purpose, interpreted, dynamic programming
language that provides constructs intended to facilitate readability and enable clear
programs on both small and large scales.
Resource
Any entity that provides a well-defined set of network functionalities that can be
modeled and controlled through an API. A resource may be an individual device or
network function, or an entire network or network domain.
27. 27
Acronyms
API Application Programming Interface
An API expresses a software component in terms of its inputs, outputs, and opera-
tions for programmatically manipulating and controlling a software component.
BP Blue Planet
Blue Planet is a Ciena software platform purpose-built for network virtualization,
orchestration, and management.
BPM Business Process Modeling
BPM is a way of expressing a business process or workflow in a human-readable
graphical form.
BPMN Business Process Modeling Notation
BPMN is a graphical notation, similar to a flow chart, for business process modeling.
BSS Business Support System
BSS is a set of software used by a telecommunications service provider to run its
business operations. Typically, BSS deals with the taking of orders, payment issues,
and revenues, and supports the management of products, orders, revenue, and
customers.
CE Carrier Ethernet
CE represents a standard set of Ethernet services that have come into being over
the past 10+ years. The standard covers point-to-point and multipoint Ethernet
connectivity services. Today, the global CE market exceeds $50 billion.
CE 2.0 Carrier Ethernet 2.0
CE 2.0 represents the latest CE certification standard. Ciena packet products are
CE 2.0 certified.
CECP Carrier Ethernet Certified Professional
CECP is a certification demonstrating expertise, skills, and knowledge of Carrier
Ethernet technologies, standards, services, and applications.
CLI Command Line Interface
CLI is a means of interacting with a computer program or networking device where-
by the user issues commands in the form of successive lines of text.
28. 28
CORD Central Office Re-architected as a Data Center
CORD represents a different way of building central offices that leverages open
source and white box technologies in favor of specialized and vendor proprietary
devices. CORD combines these open building blocks with SDN and NFV to bring
economies with the scale and agility of the cloud to service providers. CORD began
as a proof of concept sponsored by ON.Lab and AT&T. Now, companies like Ciena
are helping to bring CORD into production.
COTS Commercial Off-The-Shelf
COTS describes products and/or components that are standard manufactured
products and can be purchased readily.
DC Data Center
A DC is a facility used to house computer systems and associated components. It
includes servers, storage, networking, power, air conditioning, and security.
DCI Data Center Interconnect
DCI refers to the networking—either packet or optical—that connects data centers.
EMS Element Management System
An EMS is an application for managing network elements. An EMS typically manag-
es a single vendor’s equipment or technology.
ETSI European Telecommunications Standards Institute
ETSI is an independent, not-for-profit standardization organization in the telecom-
munications industry (equipment makers and network operators), based in Europe.
The ETSI NFV ISG produced the original white paper defining NFV and defined the
architecture for deploying NFV.
FCAPS Fault, Configuration, Accounting, Performance, and Security
FCAPS is a framework describing the major elements of network management—
fault, configuration, accounting, performance, and security.
GUI Graphical User Interface
A GUI is a graphical way of interacting with a computer program or networking de-
vice. Visualization is graphical, and commands are typically controlled with a mouse.
IaaS Infrastructure as a Service
IaaS is a form of cloud computing where compute infrastructure—that is, compute
and/or storage—is provided to an end-user as a cloud-based service.
29. 29
IP Internet Protocol
IP is the principal communications protocol in the Internet protocol suite for carrying
datagrams across a network.
MANO Management and Orchestration
MANO is the component in the NFV architecture controlling how one or more VNFs
are chained together and interconnected into an end-to-end service.
MDSO Multi-Domain Service Orchestration
MDSO is the end-to-end management and control of services over physical and
virtual networking functionality and across one or more management domains.
MPLS Multi-Protocol Label Switching
MPLS is a data encapsulation methodology used for carrying different types of
telecommunications data across a packet network. MPLS uses labels to steer traffic
from one network node to the next rather than long network addresses (which
makes MPLS more efficient than routing).
NaaS Network as a Service
NaaS is a form of cloud computing where networking and connectivity are provided
to an end-user as a service.
NE Network Element
NEs are individual networking devices being managed.
NETCONF Network Configuration Protocol
NETCONF is a network management protocol used for the configuration of network
devices.
NFV Network Functions Virtualization
NFV is a network architecture concept that uses the commercial off-the-shelf
technologies, including storage and compute, to virtualize entire classes of network
node functions used to create communication services.
NFVO Network Functions Virtualization (NFV) Orchestration
NFVO is a software component that can orchestrate the lifecycle of virtualized
network functions. This includes the creation and chaining of virtualized network
functions.
30. 30
ONF Open Networking Foundation
ONF is an organization aimed at improving networking through SDN, the OpenFlow
protocol, and related technologies.
ON.Lab Open Networking Lab
ON.Lab is an organization dedicated to developing tools and platforms and building
open source communities to realize the full potential of SDN.
ONOS Open Network Operating System
ONOS is the SDN OS for service providers. ONOS has scalability, high availability,
high performance, and abstractions that make it easy to create apps and services.
OPEN-O Open Orchestrator Project
OPEN-O is a collaborative effort to bring the industry together to develop an open
source software framework and orchestrator to enable agile SDN and NFV opera-
tions. OPEN-O was announced in February 2016.
OPNFV Open Platform for Network Functions Virtualization
OPNFV is a collaborative open platform intended to accelerate the deployment of
NFV. OPNFV is mainly focused on building NFVI and VIM.
OSM Open Source MANO
OSM is an ETSI-hosted project to develop an open source NFV MANO software
stack aligned with ETSI NFV.
OSS Operation Support System
OSS is a set of software systems used by telecommunications service providers to
manage their networks (for example, telephone networks). They support manage-
ment functions such as network inventory, service provisioning, network configura-
tion, and fault management.
PaaS Platform as a Service
PaaS is a category of cloud computing services that provides a platform allowing
customers to develop, run, and manage applications without the complexity of
building and maintaining the infrastructure typically associated with developing and
launching an app.
PCE Path Computation Element
PCE is a system component, application, or network node used to determine and
find a suitable route for connecting between a source and destination end-points.
31. 31
PNF Physical Network Function
A PNF is a physical appliance or hardware device that provides network functions.
PS Professional Services
PS are consulting services provided by a vendor to customize or fine-tune an appli-
cation or installation to suit a particular customer’s needs.
RA Resource Adapter
An RA adapts between the internal data model and an external system or resource.
REST Representational State Transfer (usage: RESTful API)
REST is an architectural style and an approach to communications often used in the
development of web services. REST is a stateless, client-server, cacheable commu-
nications protocol that, in virtually all cases, uses the HTTP protocol.
SaaS Software as a Service
SaaS is a software licensing and on-demand delivery model in which software is
licensed on a subscription basis and is centrally hosted.
SDN Software-Defined Networking
SDN is an approach to computer networking that allows network administrators to
manage network services’ higher-layer abstracted functionality. This is done by de-
coupling the control plane and data plane. The control plane is logically centralized,
and the data plane remains with forwarding devices.
SMB Small and Medium-sized Business
SMBs are businesses whose personnel numbers fall below certain limits.
SME Small and Medium-sized Enterprise
SMEs are enterprises whose personnel numbers fall below certain limits.
SNMP Simple Network Management Protocol
SNMP is a protocol for collecting and organizing information about managed
devices on IP networks and modifying that information to change device behavior.
Devices that typically support SNMP include routers, switches, servers, worksta-
tions, printers, and modem racks.
SOAP Simple Object Access Protocol
SOAP is a protocol specification for exchanging structured information in the imple-
mentation of web services in computer networks. It uses XML Information Set for
32. 32
its message format and relies on application layer protocols, most notably Hyper-
text Transfer Protocol (HTTP) or Simple Mail Transfer Protocol (SMTP) for message
negotiation and transmission.
VIM Virtual Infrastructure Manager
VIM is management software that provides centralized administration of physical
and virtual compute resources. For NFV, VIM administers the cloud resources used
to run VNFs.
VNF Virtual Network Function
A VNF is a network function that has been virtualized. A VNF is different from NFV.
VNF refers to an instance or implementation of a network function in software that is
decoupled from the underlying hardware.
VNFI VNF Infrastructure
VNFI is the compute infrastructure on which a VNF is run.
vRouter Virtual Router
vRouter is a virtualized version of router functionality.
WAN Wide Area Network
A WAN is a telecommunications network or computer network extending over a
large geographical distance—global, regional, national, or metro.
x86 Intel x86 processor architecture
x86 is a family of backward-compatible instruction-set architectures based on the
Intel 8086 CPU. x86 is commonly used to refer to commodity, commercial off-the-
shelf servers used for NFV.
XML Extensible Markup Language
XML is a human-readable markup language. XML is intended to be simple and gen-
erally useable to describe documents and arbitrary data structures. Some network
management products use XML as the protocol on the management interface.
YANG Yet Another Next Generation
YANG is a data modeling language originally created to support the NETCONF
network configuration protocol. More recently, YANG is also used for data modeling
language for a few other protocols. YANG is also sometimes used to model services.
34. 34
Abel Tong
Senior Director, Solutions Marketing,
Blue Planet
Abel Tong is Senior Director of Solutions Marketing for Ciena’s Blue Planet
software solutions. He is responsible for helping to transform networks through
the application of Software Defined Networking (SDN) and Network Function
Virtualization (NFV), to deliver value and create new services, and to simplify
network operations for Ciena’s customers.
Abel has over 15 years of networking and telecommunications systems
experience and has been an active blogger, speaker, and thought leader in the
industry. Abel joined Ciena through the acquisition of Cyan. Prior to Cyan, Abel ran
Marketing for Omnitron and led the launch of several Carrier Ethernet products.
Abel has also held positions at Aktino, Calix, ADC and Pairgain.
Abel is also a long-time contributor to the MEF. Abel leads MEF’s Project UNITE,
an industry wide collaborative initiative bringing standards development
organizations together to create the building blocks for Lifecycle Service
Orchestration and Third Network. Abel is also a member of Open Cloud Connect
(OCC), Open Daylight and the Open Networking Foundation (ONF).
35. 35
Kevin Wade
Product Marketing Team Leader,
Ciena Blue Planet
Kevin Wade is Senior Director of Product Marketing for Ciena’s Blue Planet
software portfolio. In this role, Kevin is responsible for leading the Blue Planet
product marketing team, as well as for driving the creation of programs to drive
market awareness and market share for Ciena’s industry-leading SDN/NFV
orchestration, analytics and management software solutions.
Kevin has more than 20 years of experience in the networking industry with
successful start-ups and public companies, targeting both the service provider
and enterprise markets. Kevin joined Ciena through the Cyan acquisition, where
he was responsible for the company’s product marketing and field marketing
activities.
Before joining Cyan in 2012, Kevin was Sr. Director of Product Marketing with
Force10 Networks (now Dell Force10) and also held product marketing positions
with Ascend (now part of Nokia) and Cabletron (now Extreme Networks).
An accomplished technology marketer, Kevin has presented at leading industry
conferences including Supercomm, Comptel, and Cable-Tech Expo, and has
published articles in trade magazines such as Data Center Knowledge and
Lightwave. Kevin earned his Bachelor’s degree in Finance from Northern Arizona
University in 1993.
