Parallel, Proactive and Power Efficient Virtual Network Embedding in a Green and Distributed SD-ODCN Architecture

• Main Authors: Ikhlasse Hamzaoui, Guillaume Bourgeois, Benjamin Duthil, Vincent Courboulay, Hicham Medromi
• Format: Article
• Language: English
• Published: IEEE 2021-01-01
• Series: IEEE Access
• Subjects: Parallel and proactive virtual network embedding (PPVNE); optical software defined network; power efficiency; parallel global resources topology ranking (PGTR)
• Online Access: https://ieeexplore.ieee.org/document/9369370/

With a view to overcome internet ossification problem, various Virtual Network Embedding (VNE) approaches has been proposed in last few years. Nevertheless, most of prior approaches neglect some major operational requirements implied by the inherent virtualization platforms. In the case of SD-ODCN architecture, a crucial operational requirement instance is the ability to route application-specific flows or wavelengths dynamically and efficiently across multi-tenant network providers. With the perpetual bursting of clouds and IP traffics, efficient dynamic VNE not only consists on maximizing ISPs and cloud Service Provider (SP) revenues, but also involves a strong need to reduce carbon emissions. In this article, we introduce a new Parallel, Proactive and Power Efficient VNE in a Green and Distributed SD-ODCN architecture. We first formulate a Mixed Integer Linear Programming (MILP) model purposing to maximize total intra Data Center (DC)’ servers and inter networking resources power efficiency as a function of users’ request rates. Afterward, we proposed a new green location-aware, Parallel Global resource Topology Ranking (PGTR) method, prioritizing first the greenest server and network nodes. Depending on resulted ranking process classes, a Parallel and Proactive VNE (PPVNE) is therefore proposed to effectively maximize total DC’s and networking resources power efficiency. After implementing the whole proposed algorithms namely (PGTR-PPVNE) under NSFNET network topology related data, extensive simulations results proved the improvement of the proposed (PGTR-PPVNE) approach over four other benchmark methods. More precisely, the proposed (PGTR-PPVNE) achieved 6.87% decrease, 10.77% increase, 58.54% decrease and a 1.15% increase over the proactive ACO-VNE benchmark approach, respectively in terms of total power consumption, total power efficiency, requests response times and acceptance ratio.