Moving average and neural network-based virtual machine load prediction method and system

Abstract

The invention discloses a moving average and neural network-based virtual machine load prediction method and system. The method comprises the steps of S1, collecting divided-period historical load data of a to-be-predicted time period and continuous load data before the to-be-predicted time period; S2, obtaining the divided-period historical load data of the to-be-predicted time period, and after a double moving average method is used for performing predication calculation on load inertia of a virtual machine, performing prediction to obtain a first load inertia prediction value of the next time period; S3, obtaining the continuous load data before the to-be-predicted time period, and in combination with the first load inertia prediction value, performing prediction by adopting an RBF neural network to obtain a second load inertia prediction value; and S4, taking the second load inertia prediction value as a final virtual machine load inertia prediction value and performing outputting. According to the method and the system, the hysteresis of continuous load prediction can be reduced, the timeliness of load prediction can be improved, the prediction accuracy can be improved, and the adaptive flexibility of abnormal conditions is high; and the method and the system can be widely applied to the field of virtual machine load prediction.

Description

technical field [0001] The invention relates to the field of data processing, in particular to a virtual machine load prediction method and system based on moving average and neural network. Background technique [0002] Glossary: [0003] Infrastructure as a service: IaaS, the English abbreviation, uses virtual machines, storage space, databases and other infrastructure as services, and provides them to users in the form of virtual machines; [0004] Virtual machine: refers to a complete computer system that is simulated by software and has complete hardware system functions and runs in a completely isolated environment; [0005] Load: memory usage, CPU usage, or network bandwidth usage of the virtual machine at a certain moment; [0006] Secondary moving average method: carry out a moving average of the data in the observation period from far to near according to a certain span period, and then perform a second moving average calculation on the basis of the first average...

AI Technical Summary

Problems solved by technology

The existing technology ignores the load level inertia at each specific moment, so that the prediction module cannot predict the possibility of sudden changes at the moment to be predicted in time, resulting in a certain hysteresis in the prediction

In addition, if only the time-based prediction method is used, for an abnormal day, the prediction method cannot know the abnormal situation of the day, and the prediction calculation is still performed according to the original rules, and the obtained prediction results will be separated from the actual virtual machine operation status of the day. , leading to serious over- or under-forecasting, resulting in serious waste of resources, or cloud platform turbulence due to under-forecasting

[0010] In general, the current virtual machine load prediction method has a certain lag, poor timeliness, and low prediction accuracy and flexibility

Images