The Performance Evaluate Model Of Database System Based On LQNM In Cloud Computing

Cloud computing as a business and service model, the core concept is no longer dependent on the local computer to do the calculations, but by the distribution of computing tasks in multiple computers constitute a pool of resources, so that various applications can according to the need for storage space, information services and computing capabilities. With the widespread use of cloud computing, various virtualization resources’ allocation and management based on cloud computing and the database system which as an application virtualization deployed in a virtual machine has become the focus of people. Their performance directly affects the economic benefits of cloud providers and cloud users. How to management and virtualization resources allocation to achieve effective utilization of resources, and how to configure the parameters of the database system which is deploy on virtual machine in cloud to make the performance best become a key problem.One of the key technologies to evaluate the performance of the systems is to establish its performance model. There are two methods to evaluate the system performance, they are measurement method and model method. In comparison, the model method is relatively convenient, less work and low cost, this paper choose model method for modeling. Simulation method and analysis method are two kinds of Model method, analysis method is rapid, simple and low cost. Currently, QNM and LQNM are the most commonly used of analysis methods. Because QNM can only to modeling of simple hardware resources, cannot well described hierarchical relationship between resources, so based on this extension of the LQNM. It can be modeling the software and hardware resources meantime. By analysis the relationship between the task analyses with nested call to evaluate the performance of the system, find out the performance bottleneck of system.In view of the above, this paper puts forward the virtualization layer and database system is established by combining the idea of performance model, through the analysis of operating environment, operating principles and architecture of the virtualization technologies and database system in cloud computing, we establish the LQNM of database system based on virtualization technology in cloud. The model was established under the condition of the three nodes, there were five software tasks (client, user interaction interface, system management, query processing, and data storage process) and five hardware tasks (THINK, CPU1, CPU2, DISK1, DISK2).According to the client number and load types of QNM, it can be divided into closed and opened as well as single and multi-load, using MVA to solve them is an effective way. But for LQNM, because it has no the solution of product form, it cannot be directly use the MVA to solve. SRN and MOL are used more generally so far for the method of solving the model, this article choose MOL to solve the model established. According to the MOL division method, the established model can be divided into hardware and software contention sub-models, describe the input and output parameters and get parameter values through direct or indirect method to get into the MOL and MVA algorithm to solving the system response time which the model predicted.Finally, validate the effectiveness, accuracy and versatility of the model thought experiment. Comparing the predicted and actual values of system response time, the error between them within the allowed error range (this paper assumes that the error range is10%), then the established model is effective. At the same time by comparing the results between the actual and predicted values the QNM and LQNM, we can see that LQNM model predicted values close to measured values and relatively more accurate. Because the model built in this paper was established under the condition of the three nodes, so the final paper validating the generality of the model. Comparing predicting values and actual values by using the same load in different nodes of cloud computing environment. The results are within the error range of the allowable error range, the model to the cloud computing environment of different node number is applicable.