Professional Documents
Culture Documents
Abstract— Communication networks are undergoing their next Internet transforming towards an Internet-of-Things, the notion
evolutionary step toward 5G. The 5G networks are envisioned of a customer has changed from human customers only to now
to provide a flexible, scalable, agile, and programmable network also include cars, sensors, consumer electronic items, energy
platform over which different services with varying requirements
can be deployed and managed within strict performance bounds. meters etc. With such a diverse customer base, the mobile
In order to address these challenges, a paradigm shift is taking network not only has to manage the burgeoning data volume,
place in the technologies that drive the networks, and thus but at the same time ensure that customer service requests
their architecture. Innovative concepts and techniques are being are being adequately fulfilled by the network, meeting the
developed to power the next generation mobile networks. At the respective quality-of-service or quality-of-experience require-
heart of this development lie Network Function Virtualization
and Software Defined Networking technologies, which are now ments. In order to meet the data and service requirements,
recognized as being two of the key technology enablers for the network operators are constantly expanding and upgrading
realizing 5G networks, and which have introduced a major their network infrastructure, resulting in increased capital and
change in the way network services are deployed and operated. operational expenditures (capex and opex). However, in view
For interested readers that are new to the field of SDN and NFV, of the intense competition and falling prices, the average
this paper provides an overview of both these technologies with
reference to the 5G networks. Most importantly, it describes how revenue per user is not increasing proportionately resulting
the two technologies complement each other and how they are in lower return on investment. Thus, in order to reduce costs
expected to drive the networks of near future. and increase revenue mobile networks need to take their next
Index Terms— 5G networks, NFV, SDN. evolutionary leap towards 5G, which now not only addresses
the mobile edge, but also the core network.
I. I NTRODUCTION
of addressing flexibility, agility and scalability requirements, a Service Function Chain, which determines how packets
and it leverages on the recent advances in cloud computing are forwarded from one VNF to another, to constitute a
and their support for virtualized services. On the other hand, Network Service. This already improved flexibility as well as
SDN is being developed in order to make the connectivity manageability of networks, as operators can use existing cloud
services provided by 5G networks programmable, where traffic management tools, such as Puppet, Chef, and JuJu. But at the
flows can be dynamically steered and managed in order same time, it also allowed operators to use existing and well-
to gain maximum performance benefits. However, there are known paradigms of traditional networks, like high-availability
numerous challenges for making SDN and NFV deployable concepts using redundant systems and hot-standbys.
and carrier-grade [5]. Consequently, the Open Networking However, it has been reported, e.g. in [6], that the model of
Foundation (ONF) and the ETSI NFV Industry Special Group using fat virtual machines and traditional high-availability and
have been formed to standardize various aspects of SDN and performance concepts does not translate well to the cloud.
NFV-enabled networks respectively. Simple ports of software, which was originally designed to
The subsequent sections provide an overview of the main run on specialized hardware appliances, are often not able to
technological and architectural features of NFV and SDN, deliver performance and high-availability on standard cloud
and describe how these two technologies can realize a 5G environments. For instance, cloud systems, hypervisors, and
core network. A detailed discussion on how NFV and SDN virtual machines introduce an overhead in input/output oper-
complement each other towards realizing a functional 5G core ations, which limits the performance of packet processing
network is also provided. significantly. In addition, these legacy solutions often lack
mechanisms to scale horizontally, i.e. to add more nodes to (or
II. NFV AND MANO S YSTEMS remove nodes from) a system in order to meet performance
requirements. Moreover, solutions that strive to avoid failure
Obviously, the complex architecture of upcoming 5G net- by vertical integration of failure management to an underlying
works calls for an efficient management framework that high availability platform, often fail to adapt to the cloud-
provides a uniform and coherent orchestration of various native high-availability paradigm, where service instances can
resources across the multiple layers of the 5G ecosystems. be killed and restarted any time. This is because underlying
Network Function Virtualization and their Management and assumptions and mechanisms are very different [7].
Orchestration (MANO) systems offer themselves as elegant Today’s approaches, therefore, move even further and aim
solutions, aiming at decreasing cost and complexity of imple- at a more cloud-native software design for network appli-
menting new services, maintaining running services, and cations with a much smaller footprint; not running in fat
managing available resources in existing infrastructure. Thus, virtual machines but in slim container solutions. This however,
in the following we provide a detailed introduction to NFV and imposes even more challenges on the NFV management as
MANO systems and give an overview of various open-source the number of NFV entities, which need to be orchestrated,
projects and solutions available today. increases significantly. Thus, we elaborate on management and
orchestration systems in the following.
A. Network Function Virtualization
The rise of powerful general-purpose hardware, cloud com- B. Management and Orchestration
puting technology, and flexible software defined networks, led In general, NVF Management and Orchestration systems
to the first idea of virtualizing classical network functions, such aim at a simplified handling of complex network services
as routers, firewalls, and evolved packet cores. These network using NFV technology. To this end, MANO systems have to
functions, which have been executed on dedicated and often manage virtualized infrastructure, such as cloud systems, com-
specialized hardware before, now run as software applications munication and network infrastructure, like Software Defined
in virtual machines on top of cloud infrastructure. Thus, Networks, NFV entities, like Virtualized Network Functions,
the operation of dedicated network middle-boxes transfers into and the various life-cycles of all these components. Virtu-
the operation of virtual machines and software, which paves alized Network Functions are often implemented as virtual
the way to reduce capital expenditures by using common-of- machine or container images. In view of the multi-tiered archi-
the-shelf hardware and to apply existing management practices tecture vision for 5G and the related slice concept discussed
and tools from the cloud computing space in order to automate earlier, a 5G network is mainly composed of three layers,
network operation tasks and reduce operational expenditures. namely the resource layer, the network slice instance layer and
Hope is that NFV and networked systems benefit from automa- the service instance layer. Each of these respective layers needs
tion and unified ecosystems the same way cloud environments to be managed in coordination with other layers. How these
did already. Moreover, NFV systems could embrace the high- management plane entities manage and orchestrate between
availability model of cloud systems. Rather than trying to build physical or virtual resources at their respective planes, and
an architecture that can’t fail, which is the dominant approach more importantly, how they coordinate with each to deliver
in today’s telco world, NVF aims at creating an architecture an effective 5G mobile network service platform is indeed a
that builds failure management into every part of the system challenging proposition and mandates the design and devel-
and horizontally partitions it to limit single points of failure. opment of an effective NFV Management and Orchestration
The first generation of NFV system implementations trans- system that is sensitive to the stringent carrier requirements.
ferred existing monolithic applications to big virtual machine 1) ETSI MANO Framework: The most relevant NFV
appliances, each representing a single Virtual Network Func- MANO framework today is the reference model specified
tion (VNF). Multiple VNFs are then chained together using by ETSI and depicted in Fig. 3. This framework has three
YOUSAF et al.: NFV AND SDN—KEY TECHNOLOGY ENABLERS FOR 5G NETWORKS 2471
8) Charging and billing management network domains covering all kinds of applications across
9) Network slice update/upgrade the enterprise, carrier, data-center and campus network areas.
10) VNF lifecycle management, including VNF scaling and Expanding from the initial three-layer architecture picture,
migration consisting of Infrastructure, Control and Application Layers,
11) Virtualized infrastructure management i.e., management the ONF published a detailed SDN Architecture [16], [17] that
of resource capacity, performance, fault, isolation etc. will very briefly be introduced in the following. It is based on
As mentioned earlier, the basic building block of a network the following three principles:
slice at the virtualization layer is the VNF. The MANO entity • Decoupling of control from traffic forwarding and
performs the lifecycle management of a network slice by processing. This is to enable independent deployment, life
managing the individual VNFs that are part of the network cycles and evolution of control and traffic forwarding and
slice. processing entities.
III. S OFTWARE D EFINED N ETWORKING • Logically centralized control. Logically centralized
Evidently, introducing NFV and MANO systems into 5G means that control appears from the outside, application
networks also fosters very dynamic mechanisms of data traffic perspective as a single entity, but it is not implied to be
engineering and steering. For instance, new connections have deployed in a centralized monolithic implementation.
• Programmability of network services. Interfaces between
to be set up fast and agile, e.g. to connect VNFs within and
across data centers and establish end-to-end service. Likewise, SDN components expose resource abstractions and state.
network equipment has to be updated and (re-) configured con- Applications are enabled to act on these abstractions and
tinuously to support the NFV infrastructure and architecture. states programmatically using a well defined API.
This, however, is very hard to do using traditional approaches We believe that open interfaces and related protocols,
for network operations where changes are often done (semi) like OpenFlow, are key for these systems that are built
manual at relatively long timescales, like minutes, hours, even of decoupled functional components to enable the system
days. To this end, Software Defined Networking comes into operators to deploy components from any combination of a
play to overcome the limitations of traditional networks and multitude of sources like commercial vendors and open-source
traditional network operations. groups. Open, well-defined interfaces may encourage compe-
Software Defined Networking is a network paradigm that tition between providers of community-agreed (standardized)
evolved from work done at UC Berkeley and Stanford [15]. functionality. However, proprietary features and interfaces
The motivation was to break up ossified networks by replacing should be expected to persist, especially in non-mainstream
rigid hardware-based proprietary equipment and services with areas or for specialized ad-hoc extensions.
deeply programmable common software-driven services and As shown in Fig. 5, SDN controllers are at the center of
methods that span across multiple vendor-platforms; thereby the SDN architecture as envisioned by the ONF. They are
decoupling the release cycles of agile software from the responsible for the provisioning, management and control of
comparatively slow release cycles of integrated software and services and related resources. To this end, the controller
hardware. The radical change towards making networks pro- offers so-called northbound interfaces to applications and
grammable and enabling applications and network services to southbound interfaces to the resources. Using these interfaces,
directly control the abstracted infrastructure, sparked a major users and applications have the ability to directly interact with
development direction in research and education networks and the network. Leveraging the SDN controllers northbound inter-
commercial networks, affecting especially established network face, authorized applications establish so-called management-
equipment vendors among the market players. control sessions in order to invoke services or to change the
SDN has gained a lot of traction over the past years. The state of a resources at the southbound interface. In addition,
ability to manage network services through abstractions of the administrator role is responsible to create and maintain the
lower level functionalities opens up a wide range of new environment needed to provide services to clients. It has the
architecture, management and operation options, including authority to configure the SDN controller, as well as to create
new forms of interaction between end-users applications and and manage client and server contexts. To this end, configuring
networks. Deploying agile software on white-box switches an SDN controller includes the creation of the controller itself,
is expected to improve the cost-performance behavior of the the installation and modification controller-internal policies,
network. Another great value of SDN will be the ability for and the installation and configuration of actual resources and
rapid delivery of user services while using network resources control applications.
more efficiently. Service consumption, i.e. data transfer and data processing,
takes place through the corresponding network resources.
A. OpenFlow and ONF Ultimately, user traffic is conveyed by physical resources,
An important protocol in the space of SDN is OpenFlow, which may be any number of levels of abstraction below the
which enables the communication between network infrastruc- resources visible to the client or to any particular SDN con-
ture elements and the network controlling, software-based troller. Thus, leveraging the SDN controller’s ability to abstract
entities. OpenFlow is maintained by the Open Networking complex network resources and to mediate between resources
Foundation (ONF) and today supported by all major network and control applications, also paves the way to virtualized
equipment vendors. network resources. As the controller manages the information
From ONF’s point of view, SDN has started as a vehi- flow from the resources to the applications, it can restrict the
cle to flexibly update packet forwarding algorithms. Since view and only provides a subset of resources or features to its
then, its applicability extended to the wider communications upper layers. Thus, control applications can access network
2474 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 35, NO. 11, NOVEMBER 2017
Although the community/industry is on its way and pro- [16] ONF. (2014). SDN Architecture 1.0. [Online]. Available: https://www.
gressing well to realize a large part of the 5G vision by 2020, opennetworking.org/images/stories/downloads/sdn-resource%s/
technical-reports/tr_sdn_arch_1.0_06062014.pdf
a number of research challenges and issues are still open that [17] (Feb. 2016). SDN Architecture Issue 1.1, ONF TR-521. https://
needs to be addressed in order to ensure a healthy conception www.opennetworking.org/images/stories/downloads/sdn-resources/t%
of the envisaged 5G systems. Some of those questions/issues technical-reports/TR-521_SDN_Architecture_issue_1.1.pdf
[18] “Software-defined networking: The new norm for networks,” Open
are: Netw. Found., White Paper, Apr. 2012, accessed: Nov. 2, 2017. [Online].
• How to manage/handle the agility of software, especially Available: https://www.opennetworking.org/images/stories/downloads/
in view of the trend/need of the decomposition of network sdn-resources/white-papers/wp-sdn-newnorm.pdf
[19] S. Schaller and D. Hood, “Software defined networking architecture
function into micro-functions. standardization,” Elsevier J. Comput. Standards Interfaces, vol. 54,
• How to distribute network functions onto different pp. 197–202, Nov. 2017.
[20] D. Kreutz, F. Ramos, P. E. Veríssimo, C. E. Rothenberg, S. Azodolmolky,
execution platforms, when highly programmable hard- and S. Uhlig, “Software-defined networking: A comprehensive survey,”
ware (switches, smartNICs) becomes more ubiquitous. Proc. IEEE, vol. 103, no. 1, pp. 14–76, Jan. 2015.
• How to ensure interoperability between different vendors, [21] ONF. (Apr. 2015). OpenFlow Switch Specification, Version 1.5.1,
ONF TS-025. [Online]. Available: https://www.opennetworking.org/
especially in this cloud-native environment of massively images/stories/downloads/sdn-resources/o%nf-specifications/openflow/
decomposed network functions. openflow-switch-v1.5.1.pdf
• What is the best way of translating and mapping the
[22] P. Bosshart et al., “P4: Programming protocol-independent packet
processors,” SIGCOMM Comput. Commun. Rev., vol. 44, no. 3,
customers/clients/tenants business requirements over the pp. 87–95, Jul. 2014. [Online]. Available: http://doi.acm.org/
service providers’ infrastructure. 10.1145/2656877.2656890
[23] H. Song, “Protocol-oblivious forwarding: Unleash the power of SDN
• How to ensure that the QoS and QoE requirements and
through a future-proof forwarding plane,” in Proc. 2nd ACM SIGCOMM
SLAs can be fulfilled in the cloud-native environment. Workshop Hot Topics Softw. Defined Netw., 2013, pp. 127–132. [Online].
• How to efficiently assign and manage resources for the Available: http://doi.acm.org/10.1145/2491185.2491190
[24] S. Nirasawa, M. Hara, A. Nakao, M. Oguchi, S. Yamamoto, and
multitude of slices that may exist within the same admin- S. Yamaguchi, “Network application performance improvement with
istrative domain, or traverse over different administrative deeply programmable switch,” in Proc. Adjunct 13th Int. Conf. Mobile
domains. Ubiquitous Syst., Comput. Netw. Services, 2016, pp. 263–267. [Online].
Available: http://doi.acm.org/10.1145/3004010.3004030
• How, and to what extent can the network/system man- [25] (2015). BeBa—Behavioral Based Forwarding. [Online]. Avaialble:
agement be automated in order to reduce the need for http://www.beba-project.eu
manual tasks and intervention. [26] G. Bianchi, M. Bonola, A. Capone, and C. Cascone, “OpenState: Pro-
gramming platform-independent stateful openFlow applications inside
ACKNOWLEDGMENT the switch,” SIGCOMM Comput. Commun. Rev., vol. 44, no. 2,
pp. 44–51, Apr. 2014. [Online]. Available: http://doi.acm.org/10.1145/
The authors would like to acknowledge the contributions 2602204.2602211
of their colleagues of the 5G NORMA and SONATA partner [27] R. Bifulco, J. Boite, M. Bouet, and F. Schneider, “Improving SDN with
inspired switches,” in Proc. Symp. SDN Res., 2016, pp. 11-1–11-12.
consortium, although the views expressed are those of the [Online]. Available: http://doi.acm.org/10.1145/2890955.2890962
authors and do not necessarily represent the project. [28] ONF. (Oct. 2015). Relationship of SDN and NFV, Issue 1, ONF TR-518.
[Online]. Available: https://www.opennetworking.org/ima-ges/stories/
R EFERENCES downloads/sdn-resources/techn%ical-reports/onf2015.310_Architectural
[1] (2015). Network Evolution Toward 2020 and Beyond. Accessed: _comparison.08-2.pdf
Nov. 2, 2017. [Online]. Available: http://www.nec.com/en/global/ [29] Network Function Virtualisation (NFV); Ecosystem; Report on SDN
solutions/nsp/5g_vision/doc/2020_network.pdf Usage in NFV Architectural Framework, document GS NFV-EVE
[2] NGMN Alliance. (Sep. 2016). Description of Network Slicing Con- 005 V1.1.1, ETSI NFV ISG, Dec. 2015.
cept, NGMN 5G P1, Requirements & Architecture, v1.0.8. Accessed: [30] OPNFV, “Open platform for NFV (OPNFV),” Open Source
Nov. 2, 2017. [Online]. Available: https://www.ngmn.org/uploads/media/ Project, 2017, accessed: Nov. 2, 2017. [Online]. Available:
161010_NGMN_Network_Slicing_framework_v1.0.8.pdf https://www.opnfv.org/
[3] “Study on architecture for next generation system,” 3rd Generat. Partner- [31] 5G NORMA, “5G novel radio multi-service adaptive network archi-
ship Project, Tech. Rep. TR 23.799, Dec. 2016, accessed: Nov. 2, 2017. tecture,” EU H2020 Project, 2015, accessed: Nov. 2, 2017. [Online].
[Online]. Available: http://www.3gpp.org/ftp/Specs/html-info/23799.htm Available: https://5gnorma.5g-ppp.eu/
[4] NGMN Alliance 5G White Paper, 2015. [32] 5G NORMA, “D3.2: 5G NORMA network architecture—Intermediate
[5] T. Taleb, “Toward carrier cloud: Potential, challenges, and solutions,” report,” Public Deliverable D3.2, EU H2020 Project, Jun. 2017,
IEEE Wireless Commun., vol. 21, no. 3, pp. 90–91, Jun. 2014. accessed: Nov. 2, 2017. [Online]. Available: https://5gnorma.5g-
[6] M. Taylor. (2015). The Myth of the Carrier-Grade Cloud. [Online]. ppp.eu/wp-content/uploads/2017/03/5g_norma_d3-2.pdf
Available: http://www.metaswitch.com/the-switch/the-myth-of-the- Faqir Zarrar Yousaf received the B.Sc. and
carrier-grade-cloud M.Sc. degrees in EE from the NWFP University
[7] I. Nadareishvili, R. Mitra, M. McLarty, and M. Amundsen, Microser- of Engineering and Technology, Peshawar, Pakistan,
vice Architecture: Aligning Principles, Practices, and Culture, 1st ed. in 1996 and 1999, respectively, the M.Sc. degree
Sebastopol, CA, USA: O’Reilly, 2016. in telecommunication and computers from George
[8] Network Function Virtualisation (NFV); Management and Orchestra- Washington University, USA, in 2001, and the
tion, document GS NFV-MAN 001 V1.1.1, ETSI NFV ISG, Dec. 2014. Ph.D. degree from the Dortmund University of
[9] Network Function Virtualisation (NFV); Architectural Framework,
Technology, Germany, in 2010. He is currently a
document GS NFV 002 V1.2.1, Dec. 2014.
[10] (2016). OSM: Open-Source MANO. [Online]. Available: https://osm.etsi. Senior Researcher with NEC Laboratories Europe,
org/ Heidelberg, Germany. He has extensive experience
[11] (2016). ONAP: Open Network Automation Platform. [Online]. Available: in the design development, modeling, simulation,
https://www.onap.org/ and prototyping of novel protocols, algorithms, and system methods for
[12] (2015). OpenStack Tacker. [Online]. Available: https://wiki.openstack. optimizing network performance. He has filed 11 patents while his research
org/wiki/Tacker work has been published in several peer-reviewed international journals,
[13] (2015). OpenBaton. [Online]. Available: https://openbaton.github.io/ conferences, book chapters, and Internet drafts. His current research inter-
[14] SONATA, “SONATA NFV: Agile service development and orchestration ests include NFV/SDN related technologies in the context of 5G network
in 5G virtualized networks,” EU H2020 Project, 2015, accessed: Nov. 2, architecture and operations. He is also a Delegate at the ETSI NFV standards
2017. [Online]. Available: http://sonata-nfv.eu/ organization, where he has contributed to around seven published ETSI NFV
[15] N. McKeown et al., “OpenFlow: Enabling innovation in campus net- standard documents with over 130 contributions, while being a Rapporteur for
works,” SIGCOMM Comput. Commun. Rev., vol. 38, no. 2, pp. 69–74, three active work items. He received the NECs standardization Fresh Player
2008. Award in 2015 for outstanding work in ETSI NFV.
2478 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 35, NO. 11, NOVEMBER 2017
Michael Bredel received the degree in electrical Fabian Schneider received the Ph.D. degree in
engineering from Technical University Darmstadt, computer science from TU Berlin, Germany, and
Germany, in 2007, and the Ph.D. degree from the Diploma degree from TU Mnchen, Germany.
Leibniz University Hannover, Germany, in 2012. He was with DT Labs, Berlin, where he developed
He was a Network Research Engineer with the broadband access network measurement solutions
California Institute of Technology, USA, how- and studied user behavior on the Internet. During his
ever, based at CERN, Geneva. He was a Senior post-doctoral position at Universit Pierre et Marie
Researcher with the Network Research Division, Curie, Paris, he was involved in home network
NEC Europe, Heidelberg, Germany. He is currently monitoring and network performance evaluation of
a Professor for distributed systems from the Univer- virtualized systems. He spent 12 years researching,
sity of Applied Sciences Darmstadt. His research and administrating, and designing computer networks,
development is involved in the optimization of high-speed wide area networks involving mainly on network measurement and monitoring in his scientific
and big data transfers using software-defined networking and network function career. In 2012, he joined NEC Laboratories Europe, Heidelberg, Germany,
virtualization. where he is currently a Senior Researcher. After joining NEC, he shifted
attention to network and cloud management. He is involved in research and
standardization of SDN and network management. In particular, his research
interests are identifying the right abstractions for controlling networks and on
improving scalability and performance of SDN solutions.
Sibylle Schaller received the Diploma degree in
computer science from TU Dresden, Germany. She
was the Chair of the ONF Architecture Working
Group. She was with the IBM European Networking
Center, Heidelberg, Germany, and the IBM Research
Laboratory, Haifa, Israel. In 1998, she joined NEC
Laboratories Europe, Heidelberg, where she is cur-
rently a Project Manager. Her current research inter-
ests include SDN and NFV.