A Network-Aware Optimization for VM Placement

Improving the networking data traffic in modern data centers has attracted significant importance in recent years. In contrast to existing approaches that require changes in the network architecture, this paper proposes a traffic-aware virtual machine (VM) placement for Data Center Networks (DCN). The proposed optimization strategy is able to allocate VMs according to traffic patterns. The strategy follows a Mixed Integer Linear Programming (MILP) in conjunction with networking simulation. Our approach known as MILP Flow is used to simulate the data center traffic networking. We emphasize that our approach provides the maximum throughput without losses, and optimizes better network-aware allocations than other policies in cloud networks.