Analysis Of Resource Situation Based On Cloud Monitoring Platform

In recent years,the number of cloud platform infrastructures and their available resources has increased with the rapid development of software defined network and network function virtualization technology.It is necessary to monitor and manage cloud resources in an efficient way in real-time environment,so that the cloud platform can run smoothly.At the same time,the hidden relationships among the resources should be excavated,and an accurate prediction of the development trend of cloud resources should also be made,according to the historical operational state of the cloud platform and the relevant rules of experience.It is also helpful to detect the system security threats and performance bottlenecks in time,thus providing a theoretical basis for the configuration of the security policy and the establishment of the alarm rules.This thesis makes a thorough research on the analysis of resource situation based on cloud monitoring platform.Moreover,an analysis system for resource situation based on the cloud monitoring platform has been designed by taking into consideration the following parameters: analyzing the relevant research background,including the current cloud monitoring technology,forecasting model on the state trend of cloud resource,data acquisition technology,association analysis technology and the cardinality estimation algorithm for large data.This system consists of the resource situation analysis module,the resource data acquisition module and the front interface display module.The work carried out is as follows:1.A new cardinality estimation method for streaming data is proposed,and a cardinality estimation calculator is constructed by using this method.The cardinality of target/source IP address pair at each node during a specified period of time can make an efficient,real-time and accurate statistics through this calculator.2.The ganglia data acquisition technology,custom design of the cardinality estimation calculator and the task status collector are encapsulated into an external unified data acquisition interface for cloud resource.3.A ladder architecture for cloud monitoring is constructed,in which the resource data acquisition interface act as a monitoring agent.It selects the complete types of monitoring resource data,and sets a flexible monitoring frequency for different types of resource data.4.By analyzing the characteristics of cloud platform,the corresponding forecasting model based on the task execution status at each node is established for different types of resource data,which provides an accurate real-time early warning function.5.Apriori,an association analysis algorithm is used to exploit the frequent item sets on cloud monitoring platform to analyze the hidden relations among the resources on the cloud monitoring platform accurately.6.The system supports for a real-time situation monitoring in a visual form,by displaying the status of the node and the resource usage in the cloud platform dynamically,for carrying out the corresponding management operations.The system designed in this thesis can provide for resources on the cloud monitoring platform with the functions of real-time monitoring,real-time forecasting and real-time analysis.Furthermore the system has been applied to the practical projects successfully.