Virtual machine placement algorithm based on network I/O load in HPC Cloud environment
Woosung, Wang Fei, Shi Xuan, Jin Hai, Wang Zhiwu
With the development of virtualization technology and cloud computing technology, more and more high-performance computing applications are running on cloud computing resources. In high performance computing cloud systems based on virtualization, High-performance computing applications run in multiple virtual machines that may be placed on different physical nodes. If the virtual machines of multiple traffic intensive jobs are placed on the same physical node, the network I/O resources of the physical node will be competing between the virtual machines, if the virtual machine needs more network I/O resources than the network I/O bandwidth limit of the physical node, it will seriously affect the computational performance of the traffic intensive A virtual machine placement algorithm based on network I/O load equalization (NLPA) is proposed to solve the competition problem of virtual machine to network I/O resources, which uses network I/O load equalization strategy to reduce the competition of virtual machine to network I/O resources.
Temp_12041908596297.pdf