Please use this identifier to cite or link to this item:
Title: Software-Defined Networking Based Control Flow Optimization for Multi-Cloud Environment
Authors: Chaudhary, Rajat
Supervisor: Kumar, Neeraj
Keywords: Software defined networking;Data Centers;Cloud Computing;Security and privacy
Issue Date: 24-Feb-2021
Abstract: In recent years, the huge expansion of Datacenters (DC) to execute billions of end-user applications in real-time leads to a large amount of energy consumption across the globe. So, the traditional TCP/IP-based networks which are being used for DC inter-connections are facing challenges of managing stringent Quality-of-Service (QoS) requirements of different applications of the end-users and service providers. Moreover, the existing solutions rely on distributed architecture and do not scale for large scale data centers. The issue of high power consumption at DC arises with the increase in the number of nodes and links in the network. Also, it becomes problematic on the DC whenever the underlying network resources (switches, routers) are not efficiently utilized at the time of peak data traffic resulting in high operational cost of energy utilization. However, a single controller due to its limited capacity and resources may not handle heavy load traffic generated from various smart devices. In order to handle this, multiple controllers need to be deployed at the control plane so as to ensure improved efficiency and scalability of the network. The data flow by the distributed controllers fluctuates frequently which results in an uneven load distribution amongst different controllers. Software-Defined Datacenters (SDDC) have been widely used for load-aware data management for different applications across the globe. Due to its centralized architecture, the issues of scalability along with resilience (to overcome the failure of single or multiple controllers) are still challenging because of an exponential increase in the data generated from different smart devices. Most of the solutions reported in the literature for this problem use a single controller which may not address the scalability issues. However, the issues of scalability and resilience in SDDC can be solved by deploying multiple distributed controllers at the control plane. However, the primary concern in a network having various controllers is the optimal Controller Placement Problem (CPP) to resolve the issues of fault-tolerance, latency among controllers, availability, and placement. Software-Defined Networking (SDN) emerges as one of the leading technologies to address the aforementioned issues using the programmable switches and controllers. The decoupling of control functionality from the forwarding devices to the control plane in SDN provides a unique platform to design a reconfigurable network. In the aforementioned challenges, the research work focused on three problems: (i) load balancing at the control plane, (ii) energy efficiency and fast flow forwarding for the data plane, and (iii) scalability and resilience at the control plane. This task has been accomplished in this research work with three different ii Rajat Chaudhary, 901603012 approaches. The first approach presents a Load Optimization and Anomaly Detection Scheme (LOADS) is proposed. Using LOADS, the probability of switch selection is determined using the following two factors (i) distance from the switch to the controller, and (ii) resource consumption ratio of the switch to its controller. Also, an IP flow-based network anomaly detection module has been designed to classify the traffic as malicious or normal. In order to address the network anomaly, the LOADS scheme uses Access Control Policies (ACPs) on the user’s behavior in the network. The proposed scheme is evaluated on the Mininet emulator using the POX controller with datasets of Internet Topology Zoo from the BTNorthAmerica zone. The second approach formulated the Energy-Aware Routing (EAR) problem of DCs as a Mixed Integer Non-Linear Programming (MINLP) for which an Energy-Efficient Fast Flow Forwarding (EnFlow) scheme is designed. The EnFlow scheme uses the power-saving mode of the network to solve the EAR problem. It has three modules namely– priority scheduling, routing, and re-routing. The first module works according to the First-in-First-Out Push Out Priority (FIFO-POP) scheduling using the multiple OpenFlow switches. The FIFO-POP is designed to save the energy usage of multiple switches by reducing the average waiting time of incoming packets in the queue buffers. The second module is based upon an efficient flow re-routing for a new node and link adaptation to provide the maximum bandwidth to the wired links. The third module is based upon the meta-heuristic Ant Colony Routing (ACR) to execute the stochastic decision policy on the network controller for computation of the shortest path of the forwarding nodes. The proposed EnFlow scheme is simulated using the data traces of 34 cities of NorthAmerica zone with Omnet++ 5.1 using various performance evaluation metrics. The third approach proposes Placement Availability Resilient Controller (PARC) scheme. The PARC scheme works in the following four phases: (i) stable network partitioning (ii) localization of controllers using the cooperative game theory (iii) computation of an optimal number of multiple controllers and (iv) computation of minimal extra backup controllers to improve the overall network cost. The numerical results of the PARC scheme are evaluated on Internet2 OS3E topology using POCO-toolset simulated in Matlab. Moreover, the PARC scheme outperforms the existing state-of-the-art schemes (POCO-SA, POCO-MOALO, and CNCP) for inter-controller as well as switch-to-controller latency.
Appears in Collections:Doctoral Theses@CSED

Files in This Item:
File Description SizeFormat 
RAJAT CHAUDHARY Thesis Report.pdf11.35 MBAdobe PDFThumbnail

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.