Design And Implementation Of Multi-tenant Cloud Resource Management Platform

With the rapid development of next-generation information technologies such as mobile Internet,the Internet of Things,and cloud computing,and the popularity of applications worldwide,humans have quickly entered the era of big data.More and more companies start to build high-performance and highly scalable Hadoop ecological clusters to store mine and process data.Although an enterprise can use the relevant components and services of the Hadoop ecosystem to meet its needs,in actual use,an enterprise found that because various components are isolated from each other,the installation and deployment are cumbersome,which makes the enterprise development cost higher.At the same time,different departments in the enterprise have separate clusters,inevitably causing waste of hardware resources.Moreover,the lack of unified management and monitoring of the cluster is not conducive to the maintenance of the cluster.This caused obstacles to the healthy development of the company.In response to the above problems,this paper designs and implements a multi-tenant cloud resource management platform.This platform is guided by the actual needs of an enterprise to manage the Apache Hadoop ecological cluster.The underlying layer uses Open Stack virtualization technology and introduces the concept of multi-tenancy to realize the rapid installation and deployment and monitoring and management functions of the clusters,and achieve the purpose of using hardware resources as needed.The main work of the paper is as follows:(1)Demand analysis.According to the specific usage of an enterprise,we analyze the requirements of the multi-tenant cloud resource management platform,and explain the interaction process between administrators,tenants and the platform in the platform system,and we also describe the platform's authority authentication process for administrators and tenants,and ensure data between tenants Isolation.We achieve using UML to model and explain functions such as visual quick installation and deployment,cluster monitoring,and analyzing the non-functional requirements of the platform according to the use scenario of the platform;(2)Design and implementation.Based on the platform requirements,the overall architecture of the multi-tenant cloud resource management platform is designed.First we use Mybatis to implement the data type mapping between POJO and My SQL table data in the platform system.We use Spring Security to implement tenant authentication and authorization to ensure the superior and subordinate relationships between tenants at different levels and the isolation between tenants at the same level.Quick installation and deployment of the cluster based on Open Stack is achieved,while taking into account the abnormal and fault-tolerant mechanisms in the visual installation process.Finally,the monitoring function of the cluster is completed based on Sigar.Throughout the design work,system class diagrams are used to standardize the entire coding process,and the implementation process of each function is explained in detail using diagrams;(3)System test.After completing the design and coding work,test cases for each function in the system are written,then a test environment in the laboratory according to the configuration requirements of the platform is set up,Finally we follow the steps of the test cases to test the various functions of the platform system in detail.At the same time,the stability of the system and whether the response time of each function in the system meets certain performance requirements are tested;After system testing,the platform's various functions can work well,the front-end page displays normally,and the response time of each function meets the user's expectations.The platform has sufficient functional correctness and robustness to meet the unified management of a cluster by an enterprise Demand.