Dual-Phase Resource Allocation Algorithm in Software-Defined Network SDN-Enabled Cloud

Abstract

Software Defined Networks enabled-cloud (SDN-Cloud) is experiencing rapid evolution to accommodate the explosive growth of data-driven applications. However, traditional resource allocation algorithms are encountering limitations in efficient resources management. While some existing algorithms strive to minimize power consumption, they introduce network delays, impacting overall performance. Thus, this study aims to address the prevalent challenges of performance efficiency and energy saving within distributed systems. Artificial Intelligence techniques including machine learning and fuzzy logic, are increasingly utilized to develop more adaptive and intelligent resource management models. However, given the dynamic nature of SDN-cloud environments, rapid decision-making during VM allocation is essential to prevent network delays. Furthermore, the limited computational resource of SDN controller requires cautious consideration, as extensive calculations will result in network overhead or increased power consumption. Moreover, achieving subtle balance between network performance and power efficiency still an open challenge. This research introduces Dual-Phase resource allocation Algorithm (D-Ph) for heterogeneous SDN-Cloud networks with the integration of fuzzy logic. D-Ph algorithm indicates the level of utilization of both physical and virtual machines (PM and VM) in datacenters. It aims to find the appropriate host with the necessary capabilities to meet VM resource requirements, specifically processing capacity and memory. The performance of the D-Ph algorithm is evaluated by measuring the response time, serve time of network and central processing unit (CPU), Quality of Service (QoS) violation rate, and power consumption. Results have shown distinctly that D-Ph algorithm maintain high network performance while significantly reduce total power consumption in heavy-loaded large scale network.

Keywords

• Artificial intelligence
• distributed networks
• power management
• network performance
• quality of service