Joint Placement of Virtual Machine and Routing Path Based on Virtual Extended Network

In response to a tenant's request for virtual machine usage, cloud data centers typically select a physical server for virtual machine placement and calculate the routing path between the tenant and the selected virtual machine placement server to carry traffic between them. However, it is hard to predict the impact of virtual machine placement on the network, and this sequential virtual machine placement can result in the inefficient use of network resource and congestion.To address this problem, a virtual machine and routing path joint placement algorithm, VENet, is proposed. VENet expands the original network topology into a virtual network topology by adding virtual switches and links. Based on this virtual extended network topology, the joint placement problem is approximated as a commodity flow problem, where virtual machines requests are represented as a commodity flow between the corresponding access point switch and the virtual destination switch.The commodity flow problem is modeled and solved through linear programming, obtaining the virtual machine placement location and corresponding routing path simultaneously. Experimental results show that VENet can improve the upper limit of the number of virtual machine requests that can be accepted by the data center, shorten the routing path length between the tenant and the virtual machine placement location, and reduce the network load rate of the data center.Compared to the Weighted Least Connection-Genetic Algorithm(WLC-GA), VENet reduces the routing path length and network load rate by 42% and 30%, respectively.