BMA: Bandwidth allocation management for distributed systems under cloud gaming

Cloud gaming is becoming a popular IT business among cloud services. In order to host a variety of network services, modern clouds provide great computing power by equipping with multi-core CPUs and parallel-core GPUs. In our previous work, we propose VGASA [1], a virtualized GPU resource adaptive scheduling policy in cloud gaming, solving the poor default mechanisms of GPU resource sharing. However, with the number of game player for cloud gaming rising, the user experience is inevitably limited by the network bandwidth and communication quality, due to the poor default mechanism of network resources management. To address this network limitation, we propose BAM, which combines distributed architecture and virtualized GPU resource adaptive policy in cloud gaming based on our prior work. The key module of BAM maintains a real-time monitoring of the virtual machine (VM) running status to manage the allocation of network resources, guaranteeing the better running of VMs.