Cloud computing load clustering method and system and electronic device

Abstract

The present application relates to a cloud computing load clustering method and system and an electronic device. The method comprises the following steps of (1) collecting cluster monitoring data andextracting feature vectors from the cluster monitoring data; B calculating the average contour coefficient of each feature vector, and determining the K value corresponding to each feature vector whenthe average contour coefficient is maximum; c substituting K value corresponding to each feature vector into a K-Means clustering algorithm to cluster, and obtaining the clustering results of each dimension; d combining and dividing the clustering results of each dimension to form load categories with similar characteristics. The application clusters the feature vectors based on the feature set contour coefficient estimation in a production-type cloud computing environment, and then combines all the feature vectors to form a load type with high similarity, so that the generated clusters havestronger similarity, and the accuracy of the clustering effect is ensured.

Description

technical field [0001] The present application belongs to the technical field of cloud computing, and in particular relates to a cloud computing load clustering method, system and electronic equipment. Background technique [0002] Cloud computing technology is widely used in cloud environment due to its high availability, on-demand service and low cost. The cloud environment is a distributed cluster (resource pool) composed of a large number of physical machines, enabling users to obtain computing power, storage space, and information services on demand. The key factors affecting its performance are job scheduling and resource allocation for each node. Therefore, It is extremely important for cluster management to conduct in-depth analysis of the characteristics of workloads running in the cluster, so as to allocate physical resources more reasonably and make effective decisions on the execution nodes of tasks. [0003] The cloud computing environment aggregates a large nu...

AI Technical Summary

Problems solved by technology

Moreover, most of the existing load classification methods use supervised learning methods to classify loads, and the divided load categories are several common types, without considering that with the development of technology and the diversification of people's demand for network services, the load characteristics also become more complex, and it is difficult to classify them into a specific type

Secondly, most of the current load classification models use supervised machine learning methods such as support vector machines (SVM) and statistical analysis methods for classification, while the most widely used K-Means algorithm for cluster analysis is mostly based on artificially determined The value of k is used for clustering and division, and all the selected feature vectors are used as the input of K-Means during clustering. It is impossible to measure whether the selection of K value is the best, that is, it is impossible to guarantee the object characteristics in each cluster in the divided load. have a strong similarity

Images