Design And Implementation Of IaaS Cloud Platform Based On Microservices

In recent years,more and more enterprises are using IaaS(Infrastructure as a Service)platforms as their IT infrastructure.In this process,enterprises have put forward higher and higher requirements for the easy-to-use features of the IaaS cloud platform.The easy-to-use feature not only reflects the convenience of the operation and use of the IaaS platform,but also refers to the low complexity and low workload of the operation and maintenance management of the IaaS cloud platform.In this context,designing and implementing an IaaS cloud platform that is easy to operate and use,and easy to manage and maintain has become a hot spot in the industry.This thesis aims to design and implement an IaaS cloud platform based on microservices.The cloud platform provides functions such as physical resource management,virtual resource management,operation and maintenance management,and user management.Among them,Physical resource management realizes the function of dynamic management and control of various types of infrastructure;virtual resource management realizes the access and management functions of multi-type virtual resources;operation and maintenance management realizes the function of system load analysis and anomaly detection;user management realizes the authentication and authority management functions of platform access.In this thesis,the microservice architecture is used in the IaaS cloud platform to reduce code coupling.In order to simplify the operation and maintenance of the platform,this topic proposes a multi-dimensional data detection algorithm(TS-iCutpaste:time series improved Cutpaste)based on the pre-training model,which realizes the anomaly detection of multidimensional time series data based on the pre-training model.TS-iCutpaste first designs a data preprocessing strategy to solve the problem of converting time series data to segmented data,thus ensuring the time series features and original information inside the segmented data;then the algorithm uses Resnet18 as the underlying network structure of the model,Then,the algorithm uses Gaussian density estimation function as the discriminator of sample data to realize the function of identifying anomaly points of time series data.Compared with existing methods,this method achieves higher accuracy of anomaly detection in multi-dimensional data scenarios,and can effectively overcome the network degradation problem of deep models.A series of experiments show that in the anomaly detection algorithm for multi-dimensional data,the accuracy of TS-iCutpaste is 5%higher than Cutpaste and nearly 20%higher than TS-Bert.This thesis first introduces the current research hotspots and trends of IaaS cloud platforms;then investigates the implementation technologies of different platforms,and analyzes the requirements of IaaS cloud platforms based on microservices;then proposes anomalies of pre-training for multidimensional time series data Then,the design and implementation of the IaaS cloud platform based on microservices are introduced in detail.Finally,the effectiveness of the system is verified through a series of tests.