You are on page 1of 7

Proceedings of the Seventh International Conference on Communication and Electronics Systems (ICCES 2022)

IEEE Xplore Part Number: CFP22AWO-ART; ISBN: 978-1-6654-9634-6

Quality of Services in Hybrid SDN (hSDN): A Review


Samiullah Mehraban R K Yadav
Dept. Of Computer Science and Dept. Of Computer Science and
Engineering Engineering
Delhi Technological University Delhi Technological University
2022 7th International Conference on Communication and Electronics Systems (ICCES) | 978-1-6654-9634-6/22/$31.00 ©2022 IEEE | DOI: 10.1109/ICCES54183.2022.9835774

Delhi, India Delhi, India


mehraban748@gmail.com rkyadav@dtu.nic.in

Abstract- As now a day we see that internet has become the With worldwide view and centralized network control
backbone of each and every digital society, smart devices as well Software Define Network provide reliable and flexible
as the growth of new technology and concepts like IOT, Machine network management, support advance Traffic Engineering
learning 4G, 5G and social network that totally chanced all (TE), for improvement of link unitization and network
network requirement. In the last decade Software Define throughput of the network it supports smart flow scheduling
Networking (SDN) changed the Computer Network to an system. Software Define Network is widely used in different
important and significant model as it composed of three scenario and states like Google [5], public cloud [6], NTT
different layer Data layer, Control layer and Application layer, gateway [7], and optical(IP/WDM) network [8]. Software
main Idea behind SDN is separation of Control layer and Data Define Network is used in almost every field of networking
layer and centralization of control layer to improve efficiency from cloud network [9,10] to Data Center [11-13], wide Area
and performance in the network. As Software Define N has lots
Network [14-16] to wireless network [17,18]. According to
of benefit compare to legacy network but still deployment of
pure SDN is not an easy work and it has its own limitation and
the survey Software Define Network market is expecting to
some major challenges like financial challenge, technical observe a considerable growth from 8.8 billion USD in 2018
challenge, security Issue, downtime fear of the network and to 28.9 billion USD in 2023, at a Compound Annual Growth
quality of service degradation. These challenges slow down Rate (CAGR) 26.8% during this period [19].
deployment of fully SDN and cause the birth of Hybrid Software Deployment of pure Software Define Network have its
Define Network (hSDN) which legacy devices coexist with SDN own limitation and there are some possible challenges (e.g.
devices and take advantage of both legacy and Software Define
financial challenges, technical challenges, lack of standards
Network robustness and programmability. In this paper we
for migration, security Issue, downtime fears in the network
explore a review of Quality of services in Hybrid SDN in an
approach which Quality of Service can benefit the most from the
and quality of service degradation) as Software Define
concept of Hybrid Software Define Network. Network controllers and switches are not as much stable and
reliable and OpenFlow protocol is not developed sufficiently
Keywords— SDN, hSDN, IOT QOS, 4G ,5G. [21]. These challenges hold up deployment of pure Software
Define Network and cause the birth of Hybrid Software
I. INTRODUCTION Define Network (hSDN) which Software Define Network and
There has been an unbelievable growth in the traditional devices coexist [22].
telecommunication area in the world in the earlier few years In Hybrid Software Define Network both traditional
as the development of internet is that much fast that nobody devices and Software Define Network devices work and
can believe it and has totally changed Information Technology coexist together hence no need to upgrade all of our network
requirements.as we know that the number of mobile and end devices to Software Define Networks we can upgrade our
system is rapidly increasing hence use of internet is rushing devices to Software Define Network according to our needs
via these devices. Modern computer system is packet switched and requirement, hybrid Software Define Network has many
and distributed network system which each and every digital benefits and it take advantage of both Software Define
device in the world is connected to each other with the Networks and traditional network as it take robustness of
increasing number of mobile and different devices different traditional network that tradional devices provide robust
services has created requirements for network traffic and it has routing and programmability of Software Define Network as
produced needs for flexibility and reliability in the network. Software Define Network device can emphasis on
As professional and industries face difficulties in optimization of the network, in order to provide traffic load
managing todays network and to response this difficulty balance Software Define Network devices randomly divide
design of SDN (Software Define Network) started in 2008 as traffic flows and divide these flows via multiple path, so
basic Idea behind SDN is separation of Data layer and Control network operators has the ability to manage network and
layer. In the past few years Software Define Network is among balance the load more easily. In the hybrid Software Define
one of the greatest discussed topics in the telecommunication Network Controller of network can only focus on traffic or
networks, it is a new and an important concept for network some useful flows, though lots of packets are attained by
architecture as it appeared in response to the development of robustness of traditional network protocols. At the same time,
Virtualization, mobility, Internet of Things, etc. Software the communication controls that how to control, when to
Define Networking is a developed network architecture that control and for effective and efficient working of the network
separates control layer and data layer[4]. which devices to connect [23]. different integration level is
described in detail in this study [24] like integration in control
layer [25], integration in data layer and integration in both data
and control layers [26].

978-1-6654-9634-6/22/$31.00 ©2022 IEEE 652

Authorized licensed use limited to: DELHI TECHNICAL UNIV. Downloaded on August 07,2022 at 05:22:07 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Seventh International Conference on Communication and Electronics Systems (ICCES 2022)
IEEE Xplore Part Number: CFP22AWO-ART; ISBN: 978-1-6654-9634-6

In this Paper we aim to make a picture of Quality of


Services in hybrid SDN network, related research studies are
organized as motivated in literature according to the
categories in a way which quality of service can benefit the
most from the concept of hybrid SDN network, scope of this
paper is to turn to those terms that QoS characterized by
network character like loss, delay, jitter and bandwidth with
different industries mechanisms and standards in order to
ensure high Quality performance for different application.
We have studied many paper regarding QoS in Hybrid
Software Define networking as many research in the literature
are done and proposed different methods to achieve , improve
and optimize Quality of Service in order to enhance network
performance like loss, delay, jitter and link utilization of the
network (e.g. minimizing delay, maximizing throughput ,
balancing link utilization of the network in real time network).
but in the very rare case we find that they propose a method to
increase all the parameters with good and acceptable
percentage as we will focus on maximizing throughput and Fig. 1: Software Define Network Basic Architecture
minimizing delay in Hybrid SDN network.
1. Data layer
II. SDN ARCHITECTURE
First or bottom layer is Data layer or infrastructure layer
Software Define Network is an emergent network model that consist of network devices like routers, access point,
that comes with different technologies like Network Function physical switches or virtual switches etc. it is responsible for
Virtualization and Open Flow protocols. The foremost data forwarding also collecting flow statistics and monitoring
impression behind Software Define Network architecture of network information. Open Flow protocol is the most
intended for implementation and development of agility and prevalent standard CDPI (Controller-Data Plane Interfaces)
cost effectiveness of networks function and network that is used for communication purpose among control layer
application. Main and key part behind Software Define devices and data layer devices [27-28].
Network architecture is programmability of the network that
we can easily configure and troubleshoot our network within Secure channel is the interface that connect remote
less time, overall Software Define Network is considered by controllers to data layer elements. As devices and switches are
seven important characters that is decoupling of Data layer configured and managed via controller through secure
and control layer, simple devices, central control, automation channel.
and virtualization of the network, independence vendor, 2. Control layer
programmability, and openness and the main or basic idea
behind Software Define Network is control layer Second or central layer of Software Define Network is
centralization and decoupling of Data/forwarding layer and called control layer which contain one or more than one
control layer for better performance of the network software based controller which are logically centralized, that
configure and manage data plane and carry control
Fig.1 show the basic architecture of Software Define functionality of the network. It forward logic table in data
Network that consist of three main layer that is layer and control the flow tables. Southbound interface(SBI)
Application Layer is the interface that connect control layer to data layer [29].
Basically controllers contain two important components that
Control Layer is control logic and functional elements as controllers contain
more then on function mechanisms like coordinator virtualize
Data layer etc. controller’s logic control map application requirements
of networking into orders network element resources [30].
3. Application layer
Third or last layer of Software Define Network is
application layer that have one or more than one network
application like security application, Routing Protocols,
visualization, Traffic Engineering, etc. as it hosts application
for introducing new network features, it helps control layer for
configuring network with these application requirements.
Northbound interface is the interface that used to connect
application layer with the control layer [31]. An Open Flow
switch in Software Define Network has three different parts:
Open Flow Protocol, Flow table and Secure channel

978-1-6654-9634-6/22/$31.00 ©2022 IEEE 653

Authorized licensed use limited to: DELHI TECHNICAL UNIV. Downloaded on August 07,2022 at 05:22:07 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Seventh International Conference on Communication and Electronics Systems (ICCES 2022)
IEEE Xplore Part Number: CFP22AWO-ART; ISBN: 978-1-6654-9634-6

III. HYBRID SOFTWARE DEFINE NETWORK IV. HSDN AND QOS


Hybrid Software Define Network is a networking In today’s network supporting end to end quality of
architecture which both legacy devices and Software Define services is a big issue and it is an ongoing problem so
Network devices work together or we can say that centralized Researcher have proposed different solution to solve Quality
and decentralized models coexists and communicate with of Service limitation of today’s networking models, several of
each other with different degree to control, modify, configure these solutions may failed or either not implemented properly
and managing network behavior for optimizing user
experience and network performance. as Hybrid Software If we define Quality of service it is a skill of providing
Define Network take advantage of both Software Define required services for a selected network traffic in a network as
Network and legacy Network, programmability of Software the key objective of Quality of services is providing
Define Network and robustness of traditional network, no precedence with Quality of Service restrictions (bandwidth,
need to upgrade all legacy device to Software Define loss, delay and jitter) but not limited to these characteristics.
To satisfy Quality of Services requirements for different
Network, both Traditional network device and Software
Define Network devices work together, as traditional network applications SDN provide Open control interface in order to
device provide robust routing, though Software Define support flexible network traffic scheduling as it is a perfect
Network devices work on optimization of the Network, in way to satisfy Quality of Service requirement.
traditional network legacy switches control traffic with Bi et al. [33] They proposed a scheme that is not only
distributed algorithms like IGP ( Interior Gateway Protocol) focusing on guarantees of Quality of Services of industrial
though in Software Define Network networks Controllers services but their scheme also focus on an efficient resource
route traffic based on the global view, but in case of Hybrid bandwidth utilization through traffic load balancing in
Software Define Network architecture some part of traffic will SDN/OSPF hybrid network. In order to improve QoS for
be controlled under control of legacy devices and other industrial application they propose and intelligent QoS aware
remaining part of traffic will be controlled by Software Define forwarding strategy by using SPMC (Single Path Minimum
Network controllers. Cost) forwarding scheme and K path partition algorithm for
In Hybrid SDN network traffics are forwarded based on multipath forwarding. For intelligent traffic forwarding they
decision of switching mechanism and distributed routing, design hybrid network search path (HNSP). In HSNP, they
control of traffic in Hybrid Software define network depend utilize SPMC( Single Oath Minimum Cost) forwarding if the
on various deployment strategies like (Island based, service network load is light in OSPF domain, they utilized KPF( K
based, class based, controller based, Hal based, Agent base Path Forwarding) algorithm for overloaded links to search
and overlay based) in some cases traffic control and forwarded multiple path for traffic forwarding. There result shows that
through SDN, in some cases some amount of traffic will be their approach guarantees Quality of Services requirements
controlled through SDN, others controlled with traditional for Hybrid Industrial Internet services, also they efficiently
mechanisms, decisions on how to forward and process used Open Shortest Path First(OSPF) link bandwidth and
network traffics are taken by logically centralized controllers. balancing traffic load distribution in SDN/OSPF hybrid
industrial internet.
In Hybrid Software Define Network tradional and legacy
devices still use traditional protocols like ECMP (Equal Cost Huang et al. [34] for Quality of Services optimization in
Multipath), Open Shortest Path First (OSPF) in order to hybrid Software Define Network they propose near optimal
forward packets, only Software Define Network device will traffic control method. They implemented Flow split ratio
be controlled by centralized controller. These benefits make approach via locating the Open Flow group bucket
Hybrid Software Define Network much attractive to use restrictions. With open source traffic datasets, they evaluate
though hybrid Software Define Network has its own different their proposed method, and their simulation result show that
through this method we can achieve an important
challenges that is not exists in pure Software Define Network.
improvement in network Quality of Services performance. To
According to different analyzed in recent surveys in hybrid achieve and improve QoS performance like delay, jitter, loss,
SDN network coexistence of different control plane can be and link utilization of network (minimizing delay, balancing
source of security issues, as Hybrid Software Define Network link utilization and maximizing throughput ) of the network in
is also open to wide range of security issues that inherited from real time network.
SDN network and traditional network. In real time security id
one of the main and important topic in any types of network Line et al. [35] proposed QAR(Quality of Services award
specially in SDN and hSDN network as it is open and it is adaptive routing) in a multilayer hierarchical SDN network.
centralized network and chance of attack might be increased. They used distributed hierarchical control plane architecture
Research and surveyed work in this field still need too many in large Software Define network for minimizing signaling
work and there are many challenges to be solved it need many delay through three level strategy of controllers (Super,
more practical research to solve these challenges in order to domain and slave), with the help of RL(Reinforcement
make Hybrid SDN much better and perfect for every field of Learning) and Quality of Services provisioning packet
networking. forwarding they proposes QAR( Quality of Services aware
Adaptive Routing), their result confirm that QAR provides
fast convergences and overtakes the existing learning solution
with Quality of Services provisioning facilitating practical
deployment of SDN in large scale network. QAR algorithm is
proposed by the examination of action policy, long term
revenue, quality functions and system model with reward
functions. Specially to realize efficient adaptive QoS

978-1-6654-9634-6/22/$31.00 ©2022 IEEE 654

Authorized licensed use limited to: DELHI TECHNICAL UNIV. Downloaded on August 07,2022 at 05:22:07 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Seventh International Conference on Communication and Electronics Systems (ICCES 2022)
IEEE Xplore Part Number: CFP22AWO-ART; ISBN: 978-1-6654-9634-6

provisioning routing they introduce with MDP (Markov


Decision Process) with QoS aware reward function.
Louche et al. [42] proposed geographical routing protocol
Xu et al. [36] To optimize Quality of services for for vehicular networks with clustering approachs.to send date
deployment of hybrid Software Define Network under through selecting best rely it takes three different standard:
economical constriction they propose heuristic algorithm and duration between vehicles, vehicle load, communication error
proved the approximate factor of l-le. They applied log. This approach allows selection of the most reliable
randomized rounding mechanism and depth first search vehicle by ensuring connection availability and avoiding
methor for throughput maximization. They show that both communication problems. When hybrid control layers find
simulation algorithms and analysis can obtain an important next eligible device, data layer is in charge of sending and
performance improvement and perform much better than dividing the data. there model result shows that with respect
other theoretical cases. there results show that it helps to of packet drops, throughout and average routing overhead,
improve throughput around 40% compared to other scheme there proposed scheme achieves good performance.
and there proposed routing algorithm is able to improve
throughput of the network around 31% compared to ECMP Sanchez et al. [43] They developed MILP (Mixed Integer
(Equal Cost Multipath) in Hybrid SDN network. Linear Programming) optimization Model and their aim was
to minimize routing and deployment cost and to maximized
Mondel et al. [37] they explore problem of Quality of controlled traffic in the Software Define network, as they
Services aware data flow management in existence of focus on deployments of software define network devices in
mixed flows, as this is a Network hard problem they proposed Hybrid Software Define Network incrementally. There result
game theory based mixed Data flow management structure, shows that this method improves the Quality of traffic and rate
their method has the ability of warranting 24%-47% increase that managed by Software Define Network model, as they
in throughput of the network and can reduce network delay by proposed Genetic Algorithm as an execution to achieve
77%-98%. comparable performance to the propose MILP ( Mixed Integer
Linear Programming) model, they showed that Genetic
Galan et al. [38] They solve two optimization problem Algorithm is much scalable and strong than Mixed Integer
load balance of traffic and network power consumption Linear Programming model as traffic volume rises though
reduction Their simulation result show which HSLA (Hybrid improving percentage of Software Define Network flows up
Spreading Load Algorithm) overtake other methods that to 37%.
challenge one objective only either saving of network power
consumption or traffic load balancing. Siew et al. [44] First in the networking domain they
summarize generic work flow for Machine Learning, then
Guo et al. [39] Their aim was to find best migration system from perspective of Machine Learning they study the
for the traditional routers to migrate to Software Define Problematic implementation of SDN in legacy network and
Network enabled routers that how many routers to migrate to they propose Markov Decision process and Reinforcement
SDN and to decide where to migrate, they propose heuristic Learning Techniques for the deployment problem in Software
algorithm for migration structure for the routers that gets Define Network. They propose basic workflow of Machine
maximum benefit from Traffic Engineering perspective. They Learning in networking area to formulate Hybrid Software
evaluate their algorithm by comparing to static migration Define Network deployment using Markov Decision process
algorithm and greedy migration algorithm and conducting in legacy network, further for model construction and
simulation experiments, there result shows that genetic validation Q learning or SARSA technique of reinforcement
algorithm is performing very well compared to other learning can be used.
techniques in case of properly deployed migration of 40% of
router can obtains most of the benefits. Guo et al. [45] They propose RL (Reinforcement
Learning) based approach in hybrid Software Define Network
Lin et al. [40] They propose Hybrid Software define to achieve load balancing link they address traffic changing
network architecture by using Spanning Tree Protocol as they dynamically and learns traffic splitting agent to create direct
propose Simulated Annealing base Quality of Services aware link between traffic splitting and traffic demands, as traffic
Routing (SAQR) algorithm that has the ability to change splitting agent is designed to learn offline by manipulating
weight of bandwidth and loss rate of delay to find best path
Reinforcement Learning, they emulate their work in different
for Quality of Services requirements, as they calculate their topologies and real traffic so there evaluation shows that this
projected algorithm in a hybrid Software Define Network that method can achieve similar network performance and
runs different application with multiple Quality of Services performs dominance in a fast generation routing schemes.
requirements, Their results show that this algorithm perform
much better than other related work MINA in terms of ration Table 1 shows literature of the most significant related
of delay, loss and bandwidth with 88%, 90.8% and 86.5% of work Quality of Services and Traffic Engineering (TE) in
flows meeting their respective Quality of Services Hybrid ssoftware Define Networking .
requirements.
Chienhung et al. [41] they propose hybrid Software Define
Network architecture that determine presence of traditional
network switch using Spanning Tree protocols. They allow
open flow Switch to collaborate with traditional switch
through using learning bridge protocol without requiring any
changes in traditional switches, by using this features of SDN
network Software Define Network application can find
routing path according to Quality of Service requirements and
Current network Status.

978-1-6654-9634-6/22/$31.00 ©2022 IEEE 655

Authorized licensed use limited to: DELHI TECHNICAL UNIV. Downloaded on August 07,2022 at 05:22:07 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Seventh International Conference on Communication and Electronics Systems (ICCES 2022)
IEEE Xplore Part Number: CFP22AWO-ART; ISBN: 978-1-6654-9634-6

Table 1
Traffic Engineering and Quality of Service in Hybrid SDN

Paper Title /year Method Analyze and Result Conclusion


[33] Intelligent Quality of K Path partition They proposed a scheme that is not only They proposed Quality of Services Aware Forwarding
Services aware traffic algorithm focusing on guarantees of Quality of strategy by adopting K path partition algorithm and
forwarding for Services of industrial services but their SPMC (single path minimum cost) algorithm for
SDN/OSPF industrial Single path minimum scheme also focus on and efficient resource improvement of industrial application Quality of
internet cost forwarding bandwidth utilization through traffic load Services .There result shows that their scheme
balancing in SDN/OSPF industrial internet. guarantees Quality of Services requirements for Hybrid
2020 Industrial Internet services also they efficiently used
Open Shortest Path First link bandwidth and distribution
of load balance traffic in SDN/OSPF industrial internet.

[34] Intelligent traffic Deep Reinforcement To optimize Quality of Services in hybrid with open source traffic datasets they evaluate their
control for Quality of Learning (DRL) Software Define Network they propose near proposed method and their simulation result show that
Service optimization optimal traffic control method. They through this method we can achieve an important
in hybrid SDN implemented Flow split ratio approach via improvement in network Quality of Service
locating the Open Flow group bucket performance.
2021 restrictions.

[36] Incremental Heuristic Algorithm, For deployment of hybrid Software Define They show that both simulation algorithms and analysis
deployment and depth first search Networking under economical constriction can obtain an important performance improvement and
throughput method they propose heuristic algorithm and proved perform much better than other theoretical cases. there
maximization routing the approximate factor of l-le. They applied result show that it help to improve throughput around
for hybrid SDN randomized rounding mechanism and 40% compared to other scheme and there proposed
Depth first search method for throughput routing algorithm is able to improve throughput of the
2017 maximization. network around 31% compared to ECMP(Equal cost
multipath) in Hybrid Software Define network.

[38] On the tradeoff B/W Hybrid Spreading They solve two optimization problem load Their simulation result shows that HSLA(Hybrid
load balancing and Load Algorithm balance of traffic and network power Spreading Load Algorithm) overtakes other state of the
energy efficiency in (HSLA) consumption reduction . art methods which challenge one objective only either
hybrid network saving of network power consumption or traffic load
balancing
2021

[40] QoS aware routing in Simulated annealing They propose Hybrid Software Define Their results show that this algorithm perform much
Hybrid SDN network based QoS aware network architecture by using spanning tree better rather than other related work MINA in terms of
routing algorithm protocol as they propose simulated ration of delay, loss, bandwidth with 88%,90.8% and
2017 annealing bas Quality of Service aware 86.5% of flows meeting their respective Quality of
routing (SAQR) algorithm that has the Service requirements.
ability to change weight of bandwidth and
loss rate of delay to find best path for
Quality of Service requirements, as they
calculate their projected algorithm in a
hybrid SDN Network that runs different
application with multiple Quality of
Service requirements

[43] Hybrid Incremental Mixed Integer Linear They developed MILP ( Mixed Integer There result shows that this method improve the Quality
Deployment hSDN Programming(MILP) Linear Programming) optimization Model of traffic and rate that managed by Software Define
Devices and there aim was to minimize routing and network model, as they proposed Genetic Algorithm as
Genetic Algorithm deployment cost and to maximized an execution to achieve comparable performance to the
2021 (GA) controlled traffic in the Software Define propose MILP(Mixed Integer Linear Programming)
network, as they focus deployments of model and they showed that Genetic Algorithm is much
Software Define Network devices in Hybrid more scalable and strong than the MILP(Mixed Integer
software define network incrementally. Linear Programming) model as traffic volume rises
though improving percentage of Software Define
Network Flows up to 37%.

[45] Hybrid SDN RL (Reinforcement They propose RL (Reinforcement As they emulate their work in different topologies and
deployment using Learning) Learning) based approach in hybrid real traffic so there evaluation shows that this method can
Machine learning Software Define network to achieve load achieve similar network performance and performs
balancing link they address traffic changing dominance in a fast generation routing schemes
2021 dynamically and learns traffic splitting
agent to create direct link between traffic
splitting and traffic demand, as traffic
splitting agents designed to learn offline by
manipulating Reinforcement Learning.

978-1-6654-9634-6/22/$31.00 ©2022 IEEE 656

Authorized licensed use limited to: DELHI TECHNICAL UNIV. Downloaded on August 07,2022 at 05:22:07 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Seventh International Conference on Communication and Electronics Systems (ICCES 2022)
IEEE Xplore Part Number: CFP22AWO-ART; ISBN: 978-1-6654-9634-6

[44] Hybrid SDN Reinforcement First in the networking domain they They propose basic work flow of Machine Learning in
deployment using ML learning Techniques summarize generic workflow for Machin networking area to formulate Hybrid Software define
Learning , then from the perspective of network deployment using Markov Decision process in
2021 Machine Learning they study problem of legacy network, further for model construction and
implementing of software define network in validation Q learning or SARSA technique of
tradional network, they proposed Markov reinforcement learning can be used.
Decision process and Reinforcement
learning techniques for the deployment
problem in software define network.
[39] Incremental Heuristic algorithm Their aim was to find best migration system they evaluate their algorithm by comparing to Static
Deployment for for the traditional routers to migrate to migration algorithm and greedy migration algorithm and
traffic engineering in Software define network enabled routers conducting simulation experiments, there result shows
hybrid SDN that how many routers to migrate to that genetic algorithm is performing very well compared
software define network and to decide to other techniques in case of properly deployed
2015 where to migrate, they propose heuristic migration of 40% of router can obtains most of the
algorithm for migration structure for the benefits.
routers that gets maximum benefit from
Traffic Engineering perspective.

In recent years many developing applications like video


conferencing, distance learning and many other applications
are leading the networking field in the world. Even with the REFERENCES
advantage of these Quality of Service needy application for [1] Khorsandroo, Sajad, et al. "Hybrid SDN evolution: A comprehensive
end users it still need work and it still undergo from some survey of the state-of-the-art." Computer Networks 192 (2021): 107981.
issue regarding Quality of Services or Quality of Experience [2] Sinha, Yash, and K. Haribabu. "A survey: Hybrid sdn." Journal of
requests to the clients, Quality Of Services between user and Network and Computer Applications 100 (2017): 35-55.
service providers mostly relies on Service Level Agreement [3] Huang, Xinli, et al. "A survey of deployment solutions and optimization
strategies for hybrid SDN networks." IEEE Communications Surveys &
(SLA) this method work very good for the best efforts services Tutorials 21.2 (2018): 1483-1507.
and does not support liner granular traffic control though
[4] W. Xia, Y. Wen, C. H. Foh, D. Niyato, and H. Xie, “A survey on
there are others application which there Flows are sensitive software-defined networking,” IEEE Communications Surveys & Tutorials,
to delay, bandwidth ,jitter like online gaming, VOIP and video vol. 17, no. 1, pp. 27–51, 2015.
conferencing thus requiring special handling technique for [5] S. Jain, A. Kumar, S. Mandal, J. Ong, L. Poutievski, A. Singh, S.
these types of applications. Quality of Service are basically Venkata, J. Wanderer, J. Zhou, M. Zhu, et al., “B4: Experience with a
applied into two different ways: Hard QoS and Soft QoS. globally-deployed software defined WAN,” ACM SIGCOMM Computer
Hard Quality of Service method are used to guarantees the Communication Review, vol. 43, no. 4, pp. 3–14, 2013.
Quality of Service requirement for connection but has [6] C.-Y. Hong, S. Kandula, R. Mahajan, M. Zhang, V. Gill, M. Nanduri,
and R. Wattenhofer, “Achieving high utilization with software-driven
resource limitations, Integrated Services (IntServ) is an WAN,” ACM SIGCOMM Computer Communication Review, vol. 43, no.
example of hard Quality of Services, other Quality of Service 4, pp. 15–26, 2013.
approach is Soft Quality of Service method that is not strict as [7] S. Natarajan, A. Ramaiah, and M. Mathen, “A software defined cloud-
much as hard Quality of Service method, differentiated gateway automation system using OpenFlow,” in Proc. IEEE International
services Diffserv is an example of Soft Quality of Service Conference on Cloud Networking (CloudNet), pp. 219– 226, IEEE, 2013.
method. [8] A. S. Thyagaturu, A. Mercian, M. P. McGarry, M. Reisslein, and W.
Kellerer, “Software defined optical networks (SDONs): A comprehensive
V. CONCLUSION survey,” IEEE Communications Surveys & Tutorials, vol. 18, no. 4, pp.
2738–2786, 2016.
The main problem of today network is lack of Quality of [9] Mohammad Banikazemi, David Olshefski, Anees Shaikh, John Tracey,
Services guarantee as supporting end to end support system is Guohui Wang, Meridian: an SDN platform for cloud network services, IEEE
an ongoing problem of today network, Software Define Commun. Mag. 51 (2) (2013) 120–127.
Network come to solve this problem by decoupling of control [10] Joe Mambretti, Jim Chen, Fei Yeh, Next generation clouds, the
layer and data layers that provide dynamicity and flexibility to chameleon cloud testbed, and software defined networking (sdn), in: 2015
improve network throughout and link utilization of the International Conference on Cloud Computing Research and Innovation
(ICCCRI), IEEE, 2015, pp. 73–79.
network and control layer centralization make traffic
[11] Tao Wang, Fangming Liu, Hong Xu, An efficient online algorithm for
engineering much easier task. Software Define Network has dynamic SDN controller assignment in data center networks, IEEE/ACM
lots of benefit compare to legacy network but still deployment Trans. Netw. 25 (5) (2017) 2788–2801.
of pure Software Define Network is not an easy work and it [12] L Velasco, A Asensio, JLl Berral, A Castro, Víctor López, Towards a
has its own limitation and some challenges like financial carrier SDN: An example for elastic inter-datacenter connectivity, Opt.
challenge, technical challenge, security Issue, downtime fears Express 22 (1) (2014) 55–61.
in the network and quality of service degradation. These [13] Jing Liu, Jie Li, Guochu Shou, Yihong Hu, Zhigang Guo, Wei Dai, SDN
challenges slow down deployment of fully SDN and cause the based load balancing mechanism for elephant flow in data center networks,
in: 2014 International Symposium on Wireless Personal Multimedia
birth of Hybrid Software Define Network (hSDN) which Communications (WPMC), IEEE, 2014, pp. 486–490.
legacy devices coexist with SDN devices and take advantage [14] Sushant Jain, Alok Kumar, Subhasree Mandal, Joon Ong, Leon
of both legacy and Software Define Network robustness and Poutievski, Arjun Singh, Subbaiah Venkata, Jim Wanderer, Junlan Zhou,
programmability. In this review paper we aim to make a Min Zhu, Jon Zolla, Urs Hölzle, Stephen Stuart, Amin Vahdat, B4:
picture of Quality of Services in Hybrid SDN network that Experience with a globallydeployed software defined wan, in: Proceedings
motivated in literature, we organize related research and of the ACM SIGCOMM 2013 Conference on SIGCOMM,
SIGCOMM’13,2013,pp.3–14,.
surveys studies according to categories in a way which Quality
of Service can benefit the most from the concept of Hybrid
Software Define Network.

978-1-6654-9634-6/22/$31.00 ©2022 IEEE 657

Authorized licensed use limited to: DELHI TECHNICAL UNIV. Downloaded on August 07,2022 at 05:22:07 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Seventh International Conference on Communication and Electronics Systems (ICCES 2022)
IEEE Xplore Part Number: CFP22AWO-ART; ISBN: 978-1-6654-9634-6

[15] Xin Jin, Yiran Li, Da Wei, Siming Li, Jie Gao, Lei Xu, Guangzhi Li, [36] Xu, Hongli, et al. "Incremental deployment and throughput
Wei Xu, Jennifer Rexford, Optimizing bulk transfers with software-defined maximization routing for a hybrid SDN." IEEE/ACM Transactions on
optical WAN, in: Proceedings of the 2016 ACM SIGCOMM Conference, Networking 25.3 (2017): 1861-1875.
2016, pp. 87–100. [37] A. Mondal, S. Misra, FlowMan: QoS-aware dynamic data flow
[16] Chi-Yao Hong, Subhasree Mandal, Mohammad Al-Fares, Min Zhu, management in software-defined networks, IEEE J. Sel. Areas Commun. 38
Richard Alimi, Kondapa Naidu B., Chandan Bhagat, Sourabh Jain, Jay (7) (2020) 1366–1373.
Kaimal, Shiyu Liang, Kirill Mendelev, Steve Padgett, Faro Rabe, Saikat Ray, [38] Galàn-Jimènez, Jaime, et al. "On the tradeoff between load balancing
Malveeka Tewari, Matt Tierney, Monika Zahn, Jonathan Zolla, Joon Ong, and energy-efficiency in hybrid IP/SDN networks." 2021 12th International
Amin Vahdat, B4 and after: Managing hierarchy, partitioning, and Conference on Network of the Future (NoF). IEEE, 2021.
asymmetry for availability and scale in google’s software-defined WAN, in:
Proceedings of the 2018 Conference of the ACM Special Interest Group on [39] Guo, Yingya, et al. "Incremental deployment for traffic engineering in
Data Communication, SIGCOMM ’18, 2018, pp. 74–87, hybrid SDN network." 2015 IEEE 34th international performance computing
and communications conference (IPCCC). IEEE, 2015.
[17] Salvatore Costanzo, Laura Galluccio, Giacomo Morabito, Sergio
[40] Lin, Chienhung, Kuochen Wang, and Guocin Deng. "A QoS-aware
Palazzo, Software defined wireless networks: Unbridling SDNs, in: 2012
European Workshop on Software Defined Networking, IEEE, 2012, pp.1-6. routing in SDN hybrid networks." Procedia Computer Science 110 (2017):
242-249.
[18] Roberto Riggio, Mahesh K Marina, Julius Schulz-Zander, Slawomir
[41] Chienhung Lin, Kuochen Wang, Guocin Deng, A QoS-aware routing in
Kuklinski, Tinku Rasheed, Programming abstractions for software-defined
wireless networks, IEEE Trans. Netw. Serv. Manag. 12 (2) (2015) 146–162. SDN hybrid networks, Procedia Comput. Sci. 110 (2017) 242–249, 14th
International Conference on Mobile Systems and Pervasive Computing
[19] MARKETSANDMARKETS, Software defined networking market by (MobiSPC 2017) / 12th International Conference on Future Networks and
SDN type (open SDN, SDN via overlay, and SDN via API), component Communications (FNC 2017) / Affiliated Workshops.
(solutions and services), end user (data centers, service providers, and
enterprises), and region - global forecast to 2023, 2019, [42] Lylia Alouache, Nga Nguyen, Makhlouf Aliouat, Rachid Chelouah,
HSDN-GRA: A hybrid software-defined networking-based geographic
[20] N. McKeown, T. Anderson, H. Balakrishnan, G. Parulkar, L. Peterson, routing protocol with multi-agent approach, Int. J. Commun. Syst. 33 (15)
J. Rexford, S. Shenker, and J. Turner, “OpenFlow: enabling innovation in (2020) e4521.
campus networks,” ACM SIGCOMM Computer Communication Review,
vol. 38, no. 2, pp. 69–74, 2008. [43]Cespedes-Sanchez, Pedro, et al. "Hybrid Incremental Deployment of
HSDN Devices." IEEE Transactions on Network and Service Management
[21] S. Vissicchio, L. Vanbever, and O. Bonaventure, “Opportunities and (2021).
research challenges of hybrid software defined networks,” ACM SIGCOMM
Computer Communication Review, vol. 44, no. 2, pp. 70–75, 2014. [44] Siew, Hong Wei, Saw Chin Tan, and Ching Kwang Lee. "Hybrid SDN
Deployment Using Machine Learning." Computational Science and
[22] L Velasco, A Asensio, JLl Berral, A Castro, Víctor López, Towards a Technology. Springer, Singapore, 2021. 215-225.
carrier SDN: An example for elastic inter-datacenter connectivity, Opt.
Express 22 (1) (2014) 55–61. [45] Guo, Yingya, et al. "Traffic Engineering in Hybrid Software Defined
Network via Reinforcement Learning." Journal of Network and Computer
[23] Stefano Vissicchio, Laurent Vanbever, and Olivier Bonaventure. Applications 189 (2021): 103116.
Opportunities and research challenges of hybrid software defined networks.
ACM SIGCOMM Computer Communication Review, 44(2):70–75, 2014.
[24] Yash Sinha, K Haribabu, et al. A survey: hybrid sdn. Journal of Network
and Computer Applications, 100:35 55, 2017.
[25] Matthew Caesar, Donald Caldwell, Nick Feamster, Jennifer Rexford,
Aman Shaikh, and Jacobus van der Merwe. Design and implementation of a
routing control platform. In Proceedings of the 2nd conference on
Symposium on Networked Systems Design & Implementation-Volume 2,
pages 15–28. USENIX Association, 2005.
[26] Cheng Jin, Cristian Lumezanu, Qiang Xu, Zhi-Li Zhang, and Guofei
Jiang. Telekinesis: Controlling legacy switch routing with openflow in
hybrid networks. In Proceedings of the 1st ACM SIGCOMM Symposium on
Software Defined Networking Research, pages 1–7, 2015.
[27] N. McKeown, T. Anderson, H. Balakrishnan, G. Parulkar, L. Peterson,
J. Rexford, S. Shenker, J. Turner, Openow: Enabling innovation in campus
networks, SIGCOMM Comput. Commun. Rev. 38 (2) (2008) 69-74.
[28] S. J. Vaughan-Nichols, Openow: The next generation of the network?,
IEEE Computer 44 (8) (2011) 13{15.
[29] W. Xia, Y. Wen, C.H. Foh, D. Niyato, H. Xie, A survey on software-
defined networking, IEEE Commun. Surv. Tutor. 17 (1) (2015) 27–51.
[30] SDN architecture, Tech. rep., Open Networking Foundation (ONF)
(June 2014).
[31] D. Kreutz, F. Ramos, P. Verissimo, C. Rothenberg, S. Azodolmolky,
S. Uhlig, Software-defined networking: A comprehensive survey, Proc.
IEEE 103 (1) (2015) 14–76.
[32] Mehraban, Samiullah, Komil B. Vora, and Darshan Upadhyay. "Deploy
multi protocol label switching (mpls) using virtual routing and forwarding
(vrf)." 2018 2nd International Conference on Trends in Electronics and
Informatics (ICOEI). IEEE, 2018.
[33]Bi, Yuanguo, et al. "Intelligent quality of service aware traffic
forwarding for software-defined networking/open shortest path first hybrid
industrial internet." IEEE Transactions on Industrial Informatics 16.2 (2019)
[34] Huang, Xiaohong, Man Zeng, and Kun Xie. "Intelligent traffic control
for QoS optimization in hybrid SDNs." Computer Networks 189 (2021)
[35] S. Lin, I.F. Akyildiz, P. Wang, M. Luo, QoS-aware adaptive routing in
multi-layer software defined networks: a reinforcement learning approach,
in: 2016 IEEE International Conference on Services Computing (SCC),
2016, pp. 25–33.

978-1-6654-9634-6/22/$31.00 ©2022 IEEE 658

Authorized licensed use limited to: DELHI TECHNICAL UNIV. Downloaded on August 07,2022 at 05:22:07 UTC from IEEE Xplore. Restrictions apply.

You might also like