IDEAL: Impact-Aware Virtual Data Center Embedding With Incremental Analysis for SDN-Controlled Networks

Abstract

Virtual Data Center Embedding (VDCE) maps Virtual Data Center Requests (VDCRs), comprising Virtual Machines (VMs) and Virtual Links (VLs), onto a Physical Network (PN). The problem is $\mathbb {NP}$ -hard because it requires joint allocation of computing and communication resources. In real Data Centers (DCs), VM placement is sequential and state-dependent: each accepted placement changes host utilization, energy use, and load distribution, thereby affecting subsequent mapping decisions. Existing VDCE schemes often rely on static resource snapshots or single-objective heuristics and therefore do not adequately assess the placement-wise impact on the evolving PN. This may lead to inefficient resource utilization and increased energy overhead. To address this limitation, this paper proposes Impact-Aware Virtual Data Center Embedding with Incremental Analysis for SDN-controlled Networks (IDEAL). IDEAL estimates the incremental system impact of each tentative VM placement using two criteria: marginal energy increment and incremental load-balance distortion. These criteria quantify the additional energy consumption and the change in residual-resource distribution introduced by a tentative placement. Host selection is performed using a lightweight decision-support process that combines Analytic Hierarchy Process (AHP) and Vise Kriterijumska Optimizacija I Kompromisno Resenje (VIKOR). On the evaluated 54-host spine-leaf topology, comparison with heuristic baselines shows that IDEAL reduces active-host energy consumption by 48.70%, decreases active hosts by 47.61%, and improves load-balancing efficiency by 75.72% on average. Total PN energy reduction is smaller because idle hosts retain baseline power consumption. The comparison is limited to heuristic VDCE baselines, without learning-based baseline methods.