Cloud computing resource pre-distribution achievement method

Abstract

The invention provides a cloud computing resource pre-distribution achievement method. The cloud computing resource pre-distribution achievement method detailedly analyzes design and realization of a model, helps an information technology (IT) administrator to beforehand deploy various virtual machines to satisfy unexpected requests by forecasting possible user requests, and meanwhile forecasts possible physical resource demands aiming at specific resource types so as to beforehand purchase corresponding resources. The cloud computing resource pre-distribution achievement method comprises data collection; data washing, filtering requests of users, wherein the number of times of the requests is smaller than a preset number of times; model training, building a combined forecast model based on an array of time and types of the virtual machines, and then continuously leading in washed data and carrying out the model training according to a time window until the time window is filled, and accomplishing model convergence; data forecast, forecasting resource requests of a preset time based on the trained combined forecast model; and result processing, wherein under the premise of a preset forecast time, forecast results comprise total required quantity of each type of virtual machines, required quantity of the virtual machines in the next circle, and required quantity of a specific physical resource.

Description

technical field [0001] The invention belongs to the field of cloud computing resource management, and in particular relates to a method and system for realizing preallocation of cloud computing resources. Background technique [0002] In the data center, the resource allocation of virtual machines is a basic problem for IT administrators to maintain cloud computing environments. Traditional allocation methods are basically allocated in real time based on the actual requests of users. Such allocation must have some potential problems. For example, if the entire system is in a busy state or the request is too sudden, it will cause large The processing delay; moreover, the user's resource requirements may not be met, which will greatly affect the user's satisfaction. Therefore, traditional allocation methods often require users to submit resource requests long in advance, which brings some inconvenience to users. In addition, for the purchase of physical resources, the person...

AI Technical Summary

Problems solved by technology

Traditional allocation methods are basically allocated in real time based on the actual requests of users. Such allocation must have some potential problems. For example, if the entire system is in a busy state or the request is too sudden, it will cause large The processing delay; moreover, the user's resource requirements may not be met, which will greatly affect the user's satisfaction

Therefore, traditional allocation methods often require users to make resource requests long in advance, which brings some inconvenience to users

In addition, for the purchase of physical resources, the personal experience of the administrator is mostly relied on, and there is no special mechanism to provide relevant reference information for the administrator

Images