| With the development of Internet technology and the maturity of application Software, the demand for providing the Software for the customers as a form of services is growing gradually, the study of new Software delivery technology has become the current trend, while this new pattern, SaaS (Software as a Service) is complied with the requirements of the current Software market, it can help Software developers to win customers in the form of providing services. As the core technology of SaaS applications, multi-tenant is gaining more and more attention, the idea of the multi-tenant SaaS is managing and storing data and business process of multiple tenants on the same server group of SaaS provider, its purpose is to allow multiple tenants to share software and hardware resources, improve the utilization rate of resources, reduce the average allocation on single tenant’s infrastructure and management costs.However, when the number of users that can access concurrently reach a certain one, a series of problems will be caused, the data server resources (CPU, memory, etc.) are tight, the ability of processing data can’t keep pace,and so on,and the user waiting time increases, the problem such as access error will occur, therefore, in order to achieve service level agreements (SLAs) and maintain the high performance and utilization of the applications equipment (application server and database server), solving the above problem is an important challenge for SaaS applications.Database cluster is an effective mechanism can improve the throughput and reduce the database request response time of the database, is often used to solve the network service of high data access single point bottleneck problem. But because of the blockade of foreign companies in technology, the purchase and maintenance costs are relatively high, MySQL database is welcomed by enterprises by its free and open source, use MySQL to build a database cluster system with high availability and, can effectively control the cost of IT enterprise.Based on successfully setting up a high availability MySQL cluster, this paper hope that through an in-depth study on the MySQL cluster, Comparison the performance of database cluster under different connection types; Study the effects on the database cluster performance with different number of data nodes and sql nodes combination. In found a better way of connecting to the database and the number of nodes in optimal configuration, analysis the impact of the multi-tenant combination of different business scale to the MySQL Cluster performance.This paper uses the method of test drive, after a lot of experiments, analysis the performance of database cluster under different connection types and different number of data nodes and sql nodes combination, and different scale business combinations. Analysis on the performance of MySQL cluster mainly in the CPU’s utilization rate and memeory’s utilization rate of management node (the management node also act as the application server), SQL nodes and data nodes, at the same time, analysis the response time of several typical interactions.The innovation points of this paper is mainly reflected in the:1.Analysis the performance of MySQL cluster of the keepalived and mysql-proxy connection mode,. So far, few people study the problem, some are specifically for one of them, and lack of comparative analysis.2. Study the effects on the MySQL Cluster performance with different number of data nodes and sql nodes combination. Through the experiments can be clearly seen the pressure distributed to different roles of nodes in MySQL Cluster, can have a more in-depth understanding of database cluster.3. Through the study of the effects on the MySQL Cluster performance of multi-tenants applications with different business scales, can propose suggestions aimed at overall optimization deployment of multi tenant application database, while in ensuring the good performance of data access, keep the high utilization rate of resources, which has a high practical value. |
PDF Full Text Request