Göç maliyetli sanal makine yerleştirme problemine geçici kutu paketleme tabanlı bir yaklaşım

Cloud computing is widely adopted across many industries today due to its flexibility, scalability, and efficient resource utilization. Virtual machine placement in cloud environments is a critical optimization problem that directly affects operational efficiency. An effective placement strategy both boosts performance in data centers and reduces energy and operational costs. Existing placement algorithms typically focus on dynamic resource allocation and minimizing total costs. In this study, we present a new model for the Temporal Bin Packing (TBP) problem with migration costs, which determines a placement strategy over time that incurs the minimum possible migration. The proposed model aims to minimize both VM migration and server usage costs by taking time-varying CPU demand into account. Whereas traditional models in the literature assume fixed time slots and non-migration-aware strategies, our approach provides more realistic workload management and adaptive resource allocation by handling variable-length time periods and time-dependent CPU demands. To solve the VM placement problem efficiently, we introduce T-MIG (Temporal Migration-aware Heuristic), a fast, lightweight heuristic that produces feasible, near optimal placements within seconds. Instead of assuming a fixed number of servers, T-MIG dynamically determines how many servers are needed, enhancing the model's flexibility and scalability under real-world conditions. Experimental results show that, on large-scale instances and under time constraints, T-MIG outperforms BIP solvers and offers a practical alternative for real-time decision-making systems.