Design And Implementation Of Training Platform For Database Course Based On OpenStack

With the rise of online training,some database course training platforms have emerged in recent years,and have developed into an important practical teaching tool.However,these platforms have shortcomings: firstly,they can not support new types of database training questions such as heterogeneous databases,distributed transactions and storage;secondly,they do not support the automation of the deployment of the training environment required by the questions;thirdly,the granularity of resource monitoring is at the virtual machine level,so the consumption of different database application resources can not be sensed in real time.And the exception handling capability of the user training environment is insufficient.Aiming at these deficiencies,a training platform for database courses is designed and developed.Firstly,the virtual machine created by the Open Stack platform is choosed as the carrier for the deployment of the training environment required by the questions,rewrite the Horizon component in Open Stack,and design and implement a unified virtual machine management strategy for a variety of question types.This strategy can obtain a corresponding number of virtual machines from the virtual machine resource pool according to the question type,so as to support the deployment of the training environment required for various question types including new question types.Secondly,the Ansible deployment tool is used to design automated deployment tasks for various types of training environments,and the Ansible default large file transfer process is optimized through the concurrent transmission of files in chunks.What is more,solving the dependencies between tasks by topological sorting of deployment tasks,is to realize parallel execution of deployment tasks and shorten deployment time.At the same time,the format and generation scheme of the scripts used in the initialization of various types of questions training environments are designed,so that the training environment can be automatically initialized according to the question description.Moreover,the user answering process of various question types is designed and implemented.The platform supports users to manipulate multiple nodes through jar files to complete the question type of distributed transactions,and to complete the question type of heterogeneous database through the Python programming language.Finally,by executing user answering in the constructed process,and obtaining the resource status of this process from the /proc file system,the process-level fine-grained user answering resource monitoring is realized.At the same time,the resource usage of the executing process is regularly obtained through the sampling algorithm to deal with user responses involving a large amount of data.Aiming at the abnormal user training environment,the detection and processing of abnormal virtual machine status and database abnormality are realized,and the data recovery after user network interruption and reconnection is realized.It has been verified by experiments that the training platform can support a variety of training question types.For each question type,the training platform can correctly process the user’s answers,provide monitoring results of answering resources,and can detect and restore the abnormality of the training environment such as the database abnormality.The platform can support multiple users for concurrent training,which can meet the needs of general class training.