Virtual Machine Scaling Method Considering Performance Fluctuation of Public Cloud
Yu Kaneko, Toshio Ito, Masashi Ito, Hiroshi Kawazoe · 2017
Cloud computing has been adapted for various application areas because it can reduce the time required for system development and the cost of hardware. One of the factors that degrades performance stability of applications running in the cloud is “unexpected loads”, caused by interference between Virtual Machines (VMs) coexisting on the same physical machine. In this paper, we propose a VM scaling method that forecasts performance fluctuation caused by unexpected loads to adjust the number of VMs appropriately. We evaluate the proposed method with simulation to verify that the proposed method can improve performance stability of MQTT brokers.