Which of the following best describes the role of SDN controller in networking?

In the SDN architecture, instead of having smart equipment that knows how to handle the packets, there is dumb equipment that needs to ask a logically central entity called the SDN controller what to do with the packets.

From: Tactile Internet, 2021

SSIM and ML based QoE enhancement approach in SDN context

Asma Ben Letaifa, in Advances in Computers, 2019

3.1 SDN: Software defined networking

SDN architecture presents a new solution that consists of separating the control plane from the data plane which is typically coupled together. Network functions traditionally realized in specific hardware can now be abstracted and virtualized on any equipment. A split between control and data path nodes is performed, so a centralized controller has a global view of the network while the data plane includes devices which simply forward packets following rules expressed by the controller. In order to communicate between these two layers, an open standard protocol is employed. This separation between the two layers simplifies the network management and help to simply program network control.

The main concept of SDN architecture consists on the separation between control and forwarding functions. The Figs. 1 and 2 demonstrate the multiple components of an SDN architecture which is based on three layers separated by open interfaces. The SDN application plane is a layer composed of a variety of applications that communicate via Northbound APIs. It's responsible for management, reporting functionalities such as monitoring or security. The SDN controller plane represents the main entity in the network that facilitates the creation/destruction of network paths. Typical SDN Controllers are OpenDaylight and Floodlight. The SDN data plane includes different devices deprived from any intelligence. They simply execute the controller's rules. The SDN architecture defines also the key interfaces between the different layers, which are East/West bound API that are implemented by the different controllers of the SDN and used to facilitate communications between them. Hyperflow is one representative example of such APIs. Southbound API is implemented by the different forwarding devices in the SDN and enabling the communication between these devices and the controllers. For such APIs, we can enumerate OpenFlow or NetConf. The Northbound API is Implemented by the controllers of the SDN and used to facilitate the communication between controllers and the network management applications. In such SDN architecture, enhancing the QoE became easier by implementing specific algorithms within the controller entity. The SDN manages in such case to enable new functionalities such as QoE monitoring and enforcement functions.

Which of the following best describes the role of SDN controller in networking?

Fig. 2. SDN architecture and topology.

Read full chapter

URL: https://www.sciencedirect.com/science/article/pii/S0065245819300142

Intelligent networks

Juan A. Cabrera G., ... Christof W. Fetzer, in Tactile Internet, 2021

6.2.2.1 SDN

SDN advocates replacing distributed static network protocols with centralized and flexible software applications. Legacy networks depend on hardware that implements the standardized protocols directly in the specialized microprocessors used. This means that a new release, or the deployment of a new protocol is a slow and expensive process for the network operators, because the hardware has to be changed. The main idea behind SDN is to deploy the network protocols as programmable software on the devices constituting the network infrastructure. SDN provides an Application Programmer Interface (API) for the developers to easily program the behavior of the router, switches, and other network nodes when, e.g., routing, modifying, and dropping network packets. This property enables the fast experimentation of new ideas in the deployed networks and integration and migration of security measures to arbitrary places across the network [463]. If at some point, for example, the network operator wants to test a new protocol in its infrastructure, then it does not have to go through the long-lasting processes of standardization and hardware exchange. Instead, the new protocol can be developed using the public API of the equipment, and it can be deployed as a software patch. Once the test is finished, the network operator can go back to the previous protocol by simply switching parameters in the configuration files. In the SDN architecture, instead of having smart equipment that knows how to handle the packets, there is dumb equipment that needs to ask a logically central entity called the SDN controller what to do with the packets. This controller can be implemented in a distributed fashion for resilience, but logically it is a central entity. It can be easily programmed to give instructions to the network devices on how to handle the packets and communication flows. Since the controller is a central entity, it has a overview of the whole network. Consequently, it can globally deploy optimal algorithms and protocols. SDN basically allows new functionalities to be deployed in nearly no time, relocated, and upgraded depending on the instantaneous needs of the networks. SDN offers flexibility, but due to its important role, it has become a valuable target for attackers, which requires protection [464].

Read full chapter

URL: https://www.sciencedirect.com/science/article/pii/B9780128213438000174

Smart information technology for universal healthcare

Ankur Dumka, in Healthcare Data Analytics and Management, 2019

3.2 Software Defined Networks

Innovation in data and communication has seen development of new advances in the field of information and communication technology. Technologies like online networking, distributed computing, and IoT are just a few of the latest advanced technologies. Innovation requires omnipresent availability, high data transfer capacity, and dynamic administration [18]. Software defined networks (SDNs) and network function virtualization (NFV) are new emerging system administration ideal models that change the traditional and conventional systems. These two recently emerging technologies take care of present and future prerequisites in the field of ICT. SDN is described as introducing the concept of centralization of a control plane of routers, in order to bring together basic leadership to the whole system, whereas NFV works on the concept of virtualization, to virtualize the system capacities [19].

SDN decouples the control plane from the data plane. Centralizing the control plane provides greater ease of management, control, and monitoring of the whole network from a single location. The centralization also enables global policy management of the network, event-based triggering of the network, and centralization of the information base that will help the network cope with fast-changing scenarios. Using SDN technology, all topology information can be accessed at one central location, which enables fast and easy congestion and fault detection and management by means of corrective measures, like rerouting policy management, etc. SDN is used to provide the following functionality to existing systems:

1.

Centralized Management

2.

Easily Monitored and Fault Tolerant Network

3.

Open Protocols and Interfaces

4.

Programmable Network

5.

Dynamically Configured Network

The architecture of SDN is used to distribute the tightly coupled control and forwarding plane into different layers. There are three layered reference architecture proposed by Open Network Foundation (ONF) [20] for SDN network, which is represented as below (Fig. 1):

Which of the following best describes the role of SDN controller in networking?

Fig. 1. SDN reference model.

1.

Data Layer: Data link layer is referred to as the forwarding layer, since it is used to forward the data. This layer is connected and it communicates with the control layer by means of south-bound APIs or communication interface.

2.

Control Layer: Control layer is the middle layer of architecture proposed for SDN, which is responsible for application requirement translation. This layer is also called the brain of SDN architecture, as it manages all connected open-flow elements for execution as per desired policies implemented on them.

3.

Application Layer: Application layer consists of all the applications needed for execution in the SDN network. This layer communicates with SDN applications by means of north-bound APIs or communication interface.

As mentioned, the SDN controller is called the brain of the SDN network, which takes all necessary actions for the SDN network. The nature of an SDN controller can be distributed, centralized, or hybrid. In centralized controllers, all forwarding elements are managed and coordinated by the controller itself and thus retain a global view of the SDN controller. In distributed controllers, there is more than one controller, and they are distributed over the entire network. Hybrid controllers are a combination of the centralized and distributed concept. Some of the examples of SDN controllers are FlowVisor, OpenContrail, Floodlight, OpenDaylight, Ryu, NOX FlowVisor, BEACON, and POX.

OpenFlow is the protocol commonly used in SDN networks to set up communication between controllers and switches. It is an open interface for configuring forwarding tables of network switches and routers as per the desired path of network packets derived and decided on by the SDN controller. The OpenFlow protocol uses Transmission Control Protocol (TCP) for interaction between controller and switches within the network.

Read full chapter

URL: https://www.sciencedirect.com/science/article/pii/B9780128153680000087

Internet of things for smart grid applications

Ersan Kabalci, Yasin Kabalci, in From Smart Grid to Internet of Energy, 2019

7.3.1 Software-defined networks (SDNs)

SDN provides several advances to enhance flexibility, reliability, scalability, and interoperability of IoT-based smart grid communication infrastructure as an encouraging model. It allows to dispatch control and data communication devices into different planes. Principally, SDN offers an open architecture model in three stages by separating control and data planes, enabling centralized logical control and incorporating network programming capability. Thus, SDN copes with communication problems occurred in conventional architecture combining protection control, billing, and monitoring data transmission. The open architecture of SDN prevents inefficiency of M2M communication, and provides facilitated design, deployment, management, and maintenance of communication networks. This recent networking architecture has been separated into three layers as application, control, and physical layers. The application layer performs system operations and managements as the highest layer, and control layer interfaces application layer with physical layer. The APIs performs communication between application and control layers. Thus, control layer becomes a network operating system due to APIs and manages physical layer regarding to instructions. SDN enables operators to improve network function virtualization (NFV) applications that provides aggregation of several DERs and microgrids that are using different communication technologies on a virtual network. The integration of SDN and smart grid provides several benefits for improving communication networking, real-time monitoring applications, increased latency management, and bandwidth control [40, 41].

In addition to benefits of SDN supplied to communication infrastructure, it also improves resiliency of power network against cyber-attacks. SDN enables communication system for intrusion detection, isolating selected devices upon detection and protection approach, decreasing malicious traffic and denial of service attacks, and remote control of sensors and smart meters.

Read full chapter

URL: https://www.sciencedirect.com/science/article/pii/B9780128197103000077

Software-defined networks

Justus Rischke, Hani Salah, in Computing in Communication Networks, 2020

6.3.2 SDN switches

SDN compatible switches can be implemented in hardware and software. Many manufacturers already offer OpenFlow compatible switches. The bandwidths range from Gigabit Ethernet for common business purposes with up to 64K flow table entries to 100 Gb switching capacity with 1000K table entries for edge-to-core applications [140]. In the scope of this book, we limit ourselves to software switches, which later form the basis for our ComNetsEmu emulator, introduced in Chapter 13. One of the most popular implementations for virtualized infrastructures and data centers is Open vSwitch (OVS) [141].

Read full chapter

URL: https://www.sciencedirect.com/science/article/pii/B9780128204887000189

Software-Defined Networking for the Internet of Things: Securing home networks using SDN

Shivaranjani Anbarsu, ... Vettriselvi Vetrian, in Real-Time Data Analytics for Large Scale Sensor Data, 2020

10.1 Introduction

A Software-Defined Network (SDN) is helpful in connecting smart digital appliances with members of the family in our houses today. Diversity and heterogeneity of these connected home devices are handled easily using SDN. This is because the appliances introduce protocols with various management and interoperability challenges [1]. For example, multimedia appliances have different processing powers, screen resolutions, network connection speeds, and home-networking protocols. To overcome these obstacles, various solutions have been proposed, including the provisioning of intra-home quality of service (QoS) as proposed by [2], interoperability with UniDA, and network address translation (NAT) traversal as proposed by Yoon et al [3]. However, these solutions do not cover the requirements for flexible data sharing because of prearranged networking protocols. This causes fragmentation as a major challenge in home networks. This is resolved using SDN, as it helps in configuring home devices through its control plane. It also facilitates flexible data sharing among home devices.

With SDN-enabled home platforms, device configurations and packet control are performed with ease. Otherwise these tasks are time consuming, cumbersome, and done manually. In order to configure multiple devices and to enable them to interoperate, major technical issues such as scalability, consistency, security, and privacy need to be addressed.

The interoperability issues between home consumer electronic devices in a single network were addressed by Yoon et al [3]. They proposed a tunnel-based peer-to-peer (p2p) communication scheme that guaranteed seamless and secure data exchange between mobile Internet devices behind NAT [3]. Client-/server-based architecture was used for seamless and secure p2p communication. The client module was mounted on the mobile nodes (MNs) and the server module managed the clients in real-time using the unique identifiers assigned to each client. Interoperability and management between multiple home networks were not addressed.

Boussard et al. proposed a software-based solution called SD-LAN to interconnect smart devices according to their services. In this system, SDN controls and manages parts of a smart environment through their virtual representation, which also interconnect by using SDN's architectural principles [4]. Here a home network is interconnected temporarily to the existing multihomed network. It also provided user-level privileges in accessing the network. Issues such as SD-LAN scalability, privacy, and security were not addressed.

In a smart home system, a community connector is used for integrating community services, thereby reducing the workload of the management routine [5]. A Message Queuing Telemetry Transport (MQTT) protocol was used in smart home systems to provide control services, while HTTP was used to deliver location-based information integration services [6]. The proposed solutions were used to provide various services based on the IoT context. It solely concentrated on how to integrate IoT devices in a multihomed environment to provide various services. However, issues with scalability, privacy, and security in the multihomed environment were not addressed.

Jo et al. redesigned a home router [1], and its prototype was implemented using NOX and OpenFlow to provide per-flow control. A custom Dynamic Host Configuration Protocol (DHCP) was also implemented to enable traffic isolation and accurate measurement from the IP layer [7]. The home router leveraged SDN technology to provide home users with control over the behavior of its network [7]. The user had control over their single-home network behavior but not in a multihomed network environment.

A better way to deal with security issues in home networks is to outsource the management and operation of networks to a common node that has both operations management expertise and complete view of network activity within the networks limit. As proposed by Feamster et al. [8], an offsite controller detects Internet-wide coordinated activities arising from home networks that automates necessary actions through controllable network gateways for the sake of the home user. The user need not have to worry about security issues because the third-party controller takes care of it. Appropriate balance between user privacy and security will be difficult.

Smart home networks use cloud computing technology to aid decision making. As proposed by Tom et al. [9], a risk-based integrated management of devices was followed to improve the utilization of home resources. Using automated risk assessments linked to a service-level agreement (SLA) and user/application requirements the smart home can get smarter. Cloud computing helped in managing the resources in a home network. Data privacy and manifested contracts of data consumption via sharing and usage are needed in the form of SLA and user notifications [10, 11].

One of the major issues in OpenFlow switch is its limited-size flow table, which results in displacement of flows from the flow table [8]. This work uses a hash-based placement and localized Least Recently Used (LRU)-based replacement mechanisms [12]. The algorithm is invoked only when the flow table becomes full.

The work described by Kreutz et al. [13] defines various issues related to scalability. It addresses switch design and control platforms that focus on resiliency, scalability, performance, security, and dependability. It also paved the way for newer opportunities for carrier transport networks and cloud providers. They also purview the position of SDN as a sole descriptor of the software-defined environment. Numerous SDN solutions were proposed that allowed home users to have control over their networks’ behavior.

Next, addressing the security issues in home networks, even with minimal false alarm rates, the IDS must be exceedingly capable of achieving a high attack-detection rate. To deliver accurate IDS, machine-learning and data-mining techniques, integrated with computational intelligence, are studied and tried. The use of Genetic Fuzzy Systems within a pair-wise learning were used by Elhag et al. [14] for implementing optimal IDS. This increased the benefits in two ways. The first being a fine and clear separation between the concepts, and providing a higher interpretability of the rule set through the use of fuzzy sets with linguistic labels. The second bein the divide-and-conquer learning scheme that demarcated pairs of classes that were clear contrast. This greatly improved the precision for the rare attack events, as it obtained a better separation between a “normal activity” and the different attack types. Here, a KDD CUP99 dataset was used for evaluation and testing.

Another optimal IDS using an Adaptive Neuro-Fuzzy Inference System (ANFIS) was studied and implemented [15] by Devi et al. on 5G wireless networks using a KDD CUP99 dataset for detecting an attack on the relay. The effect of modifying the membership function and machine learning algorithms are the major areas of study and analysis for accurate IDS. A device-to-device communication scenario was considered in implementing the Fuzzy Inference System (FIS), in which the end user uses relay to receive the signal. With relay in the communication network, numerous attacks have been attempted, including Masquerade, Repudiation, and Denial-of-Service (DoS) attacks, among others. These attacks are achieved by the intruder in the network attacking the relay and acquiring the user's information connected to the relay. In this work, a KDD CUP99 dataset was used for training the generated FIS model that calculates threshold value of average testing error. Among the different types of attacks in the KDD CUP99 dataset, only DoS-based attacks were considered for analyzing the proposed model.

A Bagging Ensemble method was used [16] by implementing an IDS. REPTree was used as the base class. The required feature set from NSL_KDD dataset was selected to improve classification accuracy. The experimental results showed that the Bagging Ensemble with REPTree base class achieved the highest classification accuracy. The Bagging Ensemble method took less time in building the model and provided low false positive rates compared with other machine learning techniques such as C4.5, random forest, and naïve Bayes.

Deep learning technique was applied, for flow-based anomaly detection in SDN, for this an IDS environment was applied by Tuan et al. [17]. A Deep Neural Network (DNN) model was built and trained with the NSL-KDD dataset. Only six basic features (that can be easily obtained in an SDN environment) were taken from the 41 features of the NSL-KDD dataset in training the model. The DNN model performed better than other machine learning algorithms such as J48, naïve Bayes, multilayer perceptron, random forest, and random tree [18–20].

Deep learning approaches were used in implementing a botnet IDS. Cyber security events have created awareness in harnessing the need for machine learning algorithms that adapt to evolving network systems. Signature-based approaches for identifying data types were useless in finding the attacks that took place on normal attack platforms. Human intervention proved unfit with encrypted and altered data. A neural network model can learn to identify subtle patterns in a suitably chosen input space [21–24]. A signal processing approach was introduced for classifying data files, which instantly changes to new data formats. DNN was evaluated for three input spaces consisting of the power spectral density, byte probability distribution, and sliding window entropy of the byte sequence in a file [23]. By combining all three the trained DNN was able to discriminate amongst nine common data attacks found on the Internet with 97.4% accuracy.

In a vehicular network, the deep learning approach was used in implementing an IDS. The nodes in the DNN structure are built with probability-based feature vectors from the in-vehicular network packets [25]. As compared to the traditional ANNs applied to the IDS, this technique adopts recent advances in deep learning studies, such as initializing the parameters through the unsupervised pretraining of DNNs therefore improving detection accuracy.

Read full chapter

URL: https://www.sciencedirect.com/science/article/pii/B9780128180143000103

Innovations in DCI transport networks

Loukas Paraschis, Kannan Raj, in Optical Fiber Telecommunications VII, 2020

15.5.2 Open transport model-driven networking

Complementary to the optimal system integration debate and the related hardware integration efforts, a few other, potentially more immediate, network optimization trade-offs around the optimal “SDN” transport evolution are also being increasingly debated in the context of DCI. Notably, for example, the drive for open transport, which DCI has pioneered, raises the challenge of maintaining operational simplicity in the presence of more diverse multivendor OLS architectures, for example [39]. Because SDN automation and abstraction frameworks based on model-driven networking have offered a few enabling technologies, the debate concerns how best to evolve the current model-driven networking paradigm to a higher level of abstraction. The related proposals approach network programmability “top-down” (akin to object-oriented programming); that is, the network control logic (e.g., automation or path computation) employs only abstract data model entities (links and nodes) without needing to know domain specifics (DWDM link or IP link), allowing for a separation between specifying network operators intent (what) and the system-level actuation (how). The domain specific information is known only by the systems at the nodes of the domain involved in a specific operation, based on deterministic schema translation between generic and specific data models. Such an SDN evolution (Fig. 15.15) offers two very attractive benefits: (1) Network automation is even more (ideally completely) abstracted, which improves substantially the service provisioning and availability because 70% of cloud failures are reported to happen when a network element (thus vendor specific) management operation is in progress [14], and (2) such a unified operational framework would further enable a common networkwide operating system for both WDM and packet [11], which again would substantially improve network provisioning and availability by minimizing multilayer inefficiencies (e.g., SRLG).

Which of the following best describes the role of SDN controller in networking?

Figure 15.15. The software-defined network (SDN) evolution in automation, disaggregation and abstraction of device-centric to networkwide functions [7].

While this SDN evolution vision is increasingly being adopted by most cloud and ICP network operators, however, the specific phases and implementation details are still being debated. For example, proposals regarding the most appropriate data model definitions has recently proliferated [24] beyond YANG and Openconfig [23] to a few different data modeling “languages,” such as Thrift [45] or YAML [34], practically calling for future DCI transport systems to ideally be able to provide “multilingual” vendor-neutral abstraction instantiations of their underlying domain specific data model based on programmatic representations13 of the device manageable entities [46].

Read full chapter

URL: https://www.sciencedirect.com/science/article/pii/B9780128165027000178

Machine learning methods for optical communication systems and networks

Faisal Nadeem Khan, ... Alan Pak Tao Lau, in Optical Fiber Telecommunications VII, 2020

21.6 Future role of machine learning in optical communications

The emergence of SDNs with their inherent programmability and access to enormous amount of network-related monitored data provides unprecedented opportunities for the application of ML methods in these networks. The vision of future intelligent optical networks integrates the programmability/automation functionalities of SDNs with data analytics capabilities of ML technologies to realize self-aware, self-managing, and self-healing network infrastructures. Over the past few years we have seen an increasing amount of research on the application of ML techniques in various aspects of optical communications and networking. As ML is gradually becoming a common knowledge to the photonics community, we can envisage some potential significant developments in optical networks in the near future ushered in by the application of emerging ML technologies.

Looking to the future, we can foresee a vital role played by ML-based mechanisms across several diverse functional areas in optical networks, for example, network planning and performance prediction, network maintenance and fault prevention, network resources allocation and management, etc. ML can also aid cross-layer optimization in future optical networks requiring big data analytics, since it can inherently learn and uncover hidden patterns and unknown correlations in big data, which can be extremely beneficial in solving complex network optimization problems. The ultimate objective of ML-driven next-generation optical networks will be to provide infrastructures that can monitor themselves, diagnose and resolve their problems, and provide intelligent and efficient services to the end users.

Read full chapter

URL: https://www.sciencedirect.com/science/article/pii/B9780128165027000294

Routing and Traffic Engineering in Data Center Networks

Deep Medhi, Karthik Ramasamy, in Network Routing (Second Edition), 2018

12.10 Software-Defined Networking for Data Center Networks

In Chapter 11, we present software-defined networks and discuss routing and traffic engineering in SDNs. A data center network is an appropriate place to deploy SDN due to having full control by a provider of a data center over its entire set of devices. An SDN approach typically meets the requirements discussed in Section 12.3 with a much better ability for flow control and virtual machine migration. Thus, you might wish to read Chapter 11 to see the connections with this chapter. Certainly, the traffic engineering issues discussed in Section 12.6 and Section 12.8 also apply when SDN is used in a data center network.

We point out that SDN does not mean that you must use OpenFlow (refer to Section 11.2). Certainly, an OpenFlow based SDN is used in a data center network such as Google using their own customized environment. Interestingly, you may also use SDN with BGP in a data center network. In Section 12.9, we presented how BGP is used in a data center network. You may note that there are a lot of configurations that must be coordinated to effectively use BGP in a DCN. This is where an SDN can complement BGP in effectively managing a data center network. Certainly, a question that comes to mind is why would a provider prefer SDN-BGP over SDN-OpenFlow in a data center network? It may be noted that OpenFlow is still a relatively new protocol—all switches must be OpenFlow capable. Secondly, there are a number of versions of OpenFlow—vendor implementations of different versions are still in a relatively early stage and their interoperability has not yet been fully proven. Thus, in-house expertise to deploy your own customized OpenFlow based approach is needed, as done by Google in their data centers. On the other hand, eBGP is a proven protocol with over two decades of deployment at a very large scale; vendors' implementations are also stable and interoperable. Furthermore, various features that vendors have implemented (such as Allow AS in and AS-PATH multipath relax) are suitable in a data center network. Thus, the parts on automated configuration and traffic engineering are the main places where an SDN approach can be complemented with BGP.

Briefly, an SDN approach in a BGP-based data center is helpful to accomplish the following: 1) equal-cost multipath (ECMP) for anycast prefixes for load balancing as well as changing load distribution in the event of failures and/or traffic change, and 2) configuration management and automated maintenance. For this, a controller must do functions such as topology discovery, path computation, and network state recovery. From a software point of view, an SDN controller can be implemented using REST (Representational state transfer) [273] API. Architecturally, an SDN controller will talk with all the switches as shown in Figure 12.15. The SDN controller design view is shown in Figure 12.16. In a nutshell, a BGP-SDN controller has BGP interfaces to connect to the switches with two functions: as a BGP speaker and as a BGP listener. As the BGP speaker, a route command is injected that provides the IP Prefix, NextHop, and Router ID to a switch. As a BGP listener, it receives the IP Prefix and Route ID from a switch. For this to work, four threads run that interact with RESI API for decisions. You might think that the SDN controller may look like a route reflector, but it is not since the sessions to switches from the controller are eBGP sessions; secondly, it does much more than a typical BGP speaker because of the functionalities it provides.

Which of the following best describes the role of SDN controller in networking?

Figure 12.15. SDN central controller with BGP in a data center network (controller to switch connections are shown only for a subset of switches).

Which of the following best describes the role of SDN controller in networking?

Figure 12.16. BGP-SDN controller design (source: [12,627]).

Read full chapter

URL: https://www.sciencedirect.com/science/article/pii/B9780128007372000144

Routing and Traffic Engineering in Software Defined Networks

Deep Medhi, Karthik Ramasamy, in Network Routing (Second Edition), 2018

11.6 Summary

In this chapter, we presented an overview on software defined networks. In particular, we discussed OpenFlow, a popular control communication protocol for use in SDN. A major advantage of the SDN framework is that it allows a provider to have much control over the network as opposed to IP networks where routing independently works from control. Certainly, an IP network can be traffic engineered as we discussed in Chapter 7, but the flexibility and the control you have with an SDN is much more than an IP network. To date, SDN has been widely deployed in data center networks (Routing in data center networks is discussed in Chapter 12).

We then discussed how traffic engineering for SDN networks can be either on a per-flow routing basis or an aggregated flow routing basis, and how they lend in proactive vs. reactive approaches.

We then presented a number of optimization models for traffic engineering of SDN networks, and discussed where and how they can be used for aggregated flow routing.

Read full chapter

URL: https://www.sciencedirect.com/science/article/pii/B9780128007372000132

What is the role of a controller in an SDN network?

Network Controller is the cornerstone of Software Defined Networking (SDN) management. It is a highly scalable server role that provides a centralized, programmable point of automation to manage, configure, monitor, and troubleshoot virtual network infrastructure.

Which of the following are true about SDN controller?

The SDN controller can work independently without relying any application.

Which of the following describes what the SDN control layer does to networking devices?

Which of the following describes what the SDN control layer does to networking devices that comprise the physical layer? The control layer removes the control plane from networking devices and creates a single control plane.

What is the responsibility of an SDN controller quizlet?

The SDN controller controls the data plane elements via the API. An example of such an API is OpenFlow. A switch in OpenFlow has one or more tables for packet handling rules. Each rule matches a subset of network traffic and performs actions such as dropping, forwarding, modifying etc.