The research article under consideration is Software-defined Networking (SDN): a survey written by Benzekki et al., (2017). The article reflects on how cloud computing has changed the data communication and networking for good. SDN i.e., software-designed networking is also adopted from cloud computing. SDN defines two separate segments for control decisions (network administration and management) and forwarding decisions (routers, switches and access points). This segmentation makes SDN, less complex, inexpensive and easy to manage. Authors have listed down several benefits of SDN over the classical architecture. SDN has more efficient configuration. It can be easily implemented and more flexible. Authors contend that SDN offers scalability, handling excess data load. It is more reliable as it gives notification in case of delivery failure. Explaining SDN’s profitability for customers, authors held that it is readily available when a customer requests a file or service and is highly elastic and adjust itself according to workload or demand. Nonetheless, SDN also poses some limitations like scalability challenges etc, the solutions to which have also been discussed by Benzekki et al., (2017). Since SDN is a new approach and still evolving, it still has many issues which are needed to be solved. SDN perspective keeps on changing with the evolution of SDN and issues faced by SDN controllers also change with its evolution. Thus, many challenges demand further investigation.
Take help for your assignment
Whenever you are in a need of help for your assignment, essay or report, we are here to assist you
The purpose of this paper is to analyse the article written by Benzekki et al., (2017), highlight its strengths and identify its limitations. It also explains how the study conducted by the authors on SDN paves ways for future researches in networking and data communication.
Benzenki et al., have pinpointed the challenges faced by software designed networking since its inception. This section breaks down the challenges, their solutions, advantages of the solutions and limitations of the solutions, by using primary and secondary resources.
According to the authors, as SDN has a centralized system, so as the network grows, more and more access requests are received which becomes difficult to manage by the system.
Scalability challenges can be resolved by using clustering techniques. They delegate the work overload to data plane and increase the throughput and allow smooth scalability and tradeoffs between scalability and performance to meet the requirements. Another solution is to compress the data and store it in a minimal form (Reber, 2015).
According to Timašjov (n.d.), hierarchical clustering has some complexities which makes it difficult to work well with large data sets
According to Benzekki et al., (2017), elasticity is another issue faced by cloud providers. Elasticity is of three types.
- Scalability elasticity: It ensures the growth of network according to the demand
- Elastic configuration: It is the ability of the network to change with changing demands and add, delete or edit any configuration.
- Elastic discovery: Becoming aware of new opportunities and threats and the ability to evolve according to these emerging opportunities and threats
To encounter elasticity, network distributors should use distributed controller architecture like ElastiCon. It uses dynamic adaptation of number of controllers and their location which enables SDN to grow or shrink (Reber, 2015).
Other solutions include load balancing mechanism, which is present in ElastiCon as well as FreeFlow and Hierarchical SDN model, and, includes two layers one is global and other is local. The local is activated when the load increases on global layer. In this way, load is imbalance is addressed. More agile mechanisms can be added into the network preventing disasters. It also ensures better management.
Load Balancing techniques require high investment, low reliability, high consumption and low agility (Zhong, et al., 2015).
Variance analysis should be used which is based on OpenFlow technology to monitor the traffic of the server and when the traffic becomes high, it redirects the flow to another server using OpenFlow (Zhong, et al., 2015).
Dependability is strength of a network which means that in case of any failure or faults in the network, a person can still depend upon the data produced by this source. Ensuring dependability is essential as it makes the network reliable.
Stochastic model: It evaluates the dependability (reliability and availability) of the architecture. It uses an algorithm which represents the behavior and changing operating needs.
Probabilistic model: It uses prism tool to analyze the dependability of the network.
These models analyze the reliability and availability of the network mainly focusing on control plane reliability and dependability.
Stochastic models have complex events with very few perimeters. The assumptions of these models are very rigid and there is a less chance of incorporating more instructions into them.
Incorporate adequate perimeters according to characteristics of data. Not so many to limit its descriptive power
Maintaining performance is crucial because of centralized and programmable networking. All the features of a software-defined network impact the performance of the network. Performance often slows down due to work overload issues.
Techniques like load balancing, end to end congestion control and data traffic scheduling are used for maintaining high performance. Benchmarking is used for analyzing the performance. Controllers like Beacon, Foodlight, Maestro and NOX-MT are used in this case. Network distributors should also introduce multithreading to increase response time and throughput.
The aforementioned techniques reduce the delays by eliminating the need of network rules to be asked by controller. They improve the performance and scalability, thereby increasing response time performance is increased.
Multithreading is difficult to write, debug and test. It becomes difficult to re-architecture due to end to end congestion control
Variance analysis should be used with load balancing to increase performance
According to Benzekki et al., (2017), Software-defined network due to its competencies like handling work load, ability to work under failures and attacks, make it resilient.
Use single controller in passive mode which informs other controllers. In active mode multiple controllers are used for communication.
Multihoming: Multiple providers are connected to data centers and enterprise network for increasing resilience.
Coronet: It is a program which tolerated faults in case of a failure.
Using replication, SDN will become more resilient to faults and failures. It will improve the flow and the performance. CORONET will improve the system and make it fault tolerant. It will enable the system to recover from faults more quickly.
Multihoming is a complex structure if the requirements of the customer are too many (Network design, 2016). If the requirements are complex it will create redundancy and lower the availability.
The distributors should always ask the customers to simplify their requirements
It is another challenge faced by SDN network because it does not have a built-in security system. Separating data plane with control plane eliminates some threats but it creates new loop holes. Moreover, keeping the data at centralized system also poses a threat that if there is a potential threat in system, the whole system will come down. The vulnerabilities of SDN are present at application plane, control plane and data plane. Threats of security breach can come internally, from authorized people who have access to the network or in the form of outside attacks, unstructured or conflicting rules. Structured attacks come from other networks which can lead to a serious damage. Skilled hackers can be involved in damaging the networks.
Anomaly detection software or protocols can be used to overcome security threat. The network provider can confuse the attacker by using moving target defense system. Combination of Openflow and sflow, flow stats can be analyzed to identify any anomaly. It will provide accurate anomaly detection. The combination will enable to scan the threats and detect the IP address. Moving target defense system will provide protection in case of fingerprinting, reconnaissance and service discovery. Combination of Openflow and sflow will also provide support against port scanning attacks and worm attacks.
Anomaly detection software cannot detect anomalies which are a part of normal usage. If the anomaly complies with network protocols it is never noticed. Also each protocol has to be tested and analyzed to build accuracy (Foster, 2005).
Details of normal working pattern should be made and entered in the software so that it can detect the anomaly (Foster, 2005).
To conclude, the study by Benzekki et al., (2017) has provided valuable insights about the potential advantages and limitations of software design networking. Additionally, this paper has pinpointed the limitations of the solutions proposed by the authors through primary and secondary researches. In light of the critical analysis and by evaluating extant researches on SDN, it is recommended that the future researchers should develop specific new protocols and standardize the components to avoid the inherited problems in SDN by standardization of components.
Moreover, research on control plane, the most crucial component, should be conducted and more solutions and security measures should be developed to protect SDN from failures.
Learn how can Essay Assignment Writing assist you
We are a team of professional assignment writers, essay experts, editors, proofreaders and tutors. We
can help you with all your projects, dissertations and reports. We guarantee a service that satisfies you
Benzekki. K, Fergougui. A. E, and Elalaoui A.E., (2017). Software-defined networking (SDN): a survey. Security Comm. Networks, pp. 9:5803–5833. Doi: 10.1002/sec.1737
Foster, J., 2005. IDS: Signature versus anomaly detection. [Online]
Available at: http://searchsecurity.techtarget.com/tip/IDS-Signature-versus-anomaly-detection
H, P. M. & Rajani, I. K., 2015. Improve Performance of clustering on Cloud Datasets using improved Agglomerative CURE Hierarchical Algorithm, s.l.: s.n.
Netwrok design, 2016. Multihoming the Internet Connection. [Online]
Available at: https://www.ccexpert.us/network-design-2/multihoming-the-internet-connection.html
Reber, A., 2015. On the Scalability of the Controller in Software-Defined Networking, s.l.: s.n.
Timašjov, D., n.d. Evaluating Clustering Techniques, s.l.: s.n.
Zhong, et al., 2015. An Efficient SDN Load Balancing Scheme Based on Variance Analysis for Massive Mobile Users. Mobile Information Systems, 6 December.