Method for selecting to-be-increased components based on improved multi-target particle swam optimization algorithm

Abstract

The invention discloses a method for selecting to-be-increased components based on an improved multi-target particle swam optimization algorithm, and belongs to the technical field of cloud service optimization. The method comprises the following steps of analyzing concurrent quantity of history application users, and utilizing an autocorrelation coefficient analysis method to divide the concurrent quantity into a stationary type and a non-stationary type according to the data distribution trend of the concurrent quantity; and predicating the concurrent quantity of application users for different characteristics by utilizing different time sequence prediction methods. According to the method, the prediction efficiency is not only improved, but the flexibility of the prediction method is improved; furthermore, the component call relationship and the component call frequency are utilized to decompose the concurrent quantity of the application users, the concurrent quantity of each component is computed, two factors capable of directly reflecting the working performance of each component, namely the concurrent quantity of each component and the component response time, are comprehensively considered and are taken as the basis of selection of the to-be-increased components, so that the accuracy of the method for selecting the component is improved; the method aims at the problem of low precision of the constraint condition of the conventional particle swam optimization algorithm, and provides a semi-feasible region, so that the precision of the constraint condition is effectively improved.

Description

technical field [0001] The invention belongs to the technical field of cloud service optimization, and in particular relates to a method for selecting components to be added based on an improved multi-objective particle swarm optimization algorithm. Background technique [0002] With the complexity of the cloud service system and the openness, dynamics and uncontrollability of the operating environment, the importance of self-optimization and adaptive capabilities of the service system has become increasingly prominent; the cloud service system is realized by multiple components working together. The component service is deployed in the cloud virtual resource pool as the basic unit. The cloud service performance adaptive optimization method satisfies the characteristics of cloud computing utility computing and pay-as-you-go, so that the cloud service system can not only keep the cost of resources at a minimum To meet the dynamic resource configuration required by the applica...

AI Technical Summary

Problems solved by technology

[0005] (1) In terms of predicting the number of concurrent users using time series algorithms, the data of concurrent application users is generally considered to be stable by default, and the single exponential smoothing method is used to predict the trend of concurrent users of applications in a certain period of time in the future. However, in actual cloud service systems, different During the period of time, the concurrency of application users may belong to different data characteristics, and the prediction accuracy of only a single model is low;

[0006] (2) In terms of component selection, the current method is only based on the single performance of the component, such as component call frequency, response time, cohesion and coupling relationship to select the component to be added, and the factors that affect the work of the component are the result of a combination of various factors, only One factor cannot accurately select the components to be added;

[0007] (3) In solving the selection results of the components to be added, the traditional particle swarm optimization algorithm mainly uses the penalty function method to deal with the constraint conditions. These penalty coefficients can only be calculated with the basic experimental data, and there is a problem of low accuracy of constraint conditions

Images