The most important metrics for measuring the performance of a database system are queries per second (QPS)/transactions per second (TPS) and response time (RT). QPS/TPS refers to the number of requests completed within one second, while RT refers to the time taken to complete a single request. Improving QPS/TPS allows for better utilization of server resources, cost-effectiveness, and lower total cost of ownership (TCO) for the database server. Reducing RT not only enhances the QPS of upstream application systems and improves response speed and user experience but also further enhances the QPS/TPS of the database system.
Performance tuning, also known as performance optimization, focuses on improving QPS/TPS and reducing RT through various techniques. Performance tuning usually involves three steps: determining optimization directions, identifying bottlenecks, and finding suitable optimization solutions. It is a systematic process that requires a comprehensive understanding of hardware, operating systems, and software, along with proficiency in kernel principles and tuning tools.
OceanBase Database, as a distributed database, has been introduced with its high scalability and availability features in previous chapters. In this chapter, we will delve into the realm of high-performance and explore the performance tuning techniques of OceanBase Database.