A Taxonomy of Load Balancing Mechanisms in Centralized and Distributed SDN Architectures
- PDF / 674,908 Bytes
- 16 Pages / 595.276 x 790.866 pts Page_size
- 100 Downloads / 167 Views
SURVEY ARTICLE
A Taxonomy of Load Balancing Mechanisms in Centralized and Distributed SDN Architectures Farah Chahlaoui1 · Hamza Dahmouni1 Received: 24 May 2020 / Accepted: 5 August 2020 © Springer Nature Singapore Pte Ltd 2020
Abstract Recent Research have shown that software-defined networking (SDN) drastically improves network resource utilization, manages today’s complex networks and reduces cost. The rapid development of technology and the explosively growing information services as well as the increasing number of users, have made the load distribution a huge issue that could eventually impact the network’s performance and its quality of service. The need for SDN comes from the load-balancing problematic which could reduce the availability of network applications, resource utilization and QoS indicators such as latency, throughput and response time. In this paper, we analyze the impact of software-defined networks design and the architecture of its control plane on the load-balancing methods efficiency. We present a detailed analysis of some load-balancing algorithms and metrics in centralized and distributed SDN architectures. We also introduce some of the load-balancing mechanisms applied in 5G networks in particular. Keywords Load balancing · Software defined networks · OpenFlow · SDN architecture · 5G · QoS
Introduction Software defined networking [1, 2] provide a centralized network and trafic management. The network administrators can program forwarding decision calculation by using the OpenFlow protocol [3] in a centralized manner. Its general view of the network, its programmability and cooperativeness enables SDN with a better management of the traffic flows in the network taking into consideration their characteristics and quality of service requirements. SDN also provides the network with a better flexibility and reactivity to the network’s unexpected changes. Load balancing issue in SDN is still not thoroughly studied and more investigations should be done on this area of the field. The ultimate objective is to reactively indicate to the incoming trafic the best transmission paths in the network with the highest quality of service possible.
* Farah Chahlaoui [email protected] Hamza Dahmouni [email protected] 1
Institut National des Postes et Télécommunications, INPT, 2, av. Allal El Fassi, Madinat Al Irfane, 10000 Rabat, Morocco
Compared to traditional networks, software-defined networks (SDN) paradigm [4] have many advantages including the efficiency and simplicity of network management and security policies application. SDN overcomes the traditional networks short-comings that includes limited and inaccurate view of the network causing their inability to achieve end-toend quality of service and optimized load balancing. The architecture of the controller has a great impact on the performance of SDN. Numerous controller architectures have been proposed in the body of research, among them the centralized controller, the distributed but logically centralized controller and the comp
Data Loading...
