Research on Database Parameters Tuning Method Based on Embedded Device
Xiaoli Geng, Wenchao Xu, Yonghong Yin · Journal of Physics Conference Series · 2021
Abstract Various kinds of embedded terminal devices have enriched our lives. As a consequence, a huge amount of data has been generated, which often results in slow operation of the devices and a dramatic drop in data processing capacity. Database parameter tuning will be an important way to maintain or improve the performance of the devices. As database performance is affected by multiple parameters, manual adjustment is more and more difficult. With the development of machine learning, database automatic parameter tuning technology has become one of the main choices to solve this problem. This paper proposes to use MARS regression algorithm to optimize the database parameters of embedded devices, to perform regression prediction and fitting on each part of the data space respectively, and obtain the optimal parameter through the processes of forward, backward, and model selection. Experiments show that compared with the traditional method, the parameter optimization results generated through the method proposed in this article are very nearly the same as those obtained through the traditional method, but the time efficiency is significantly improved, hence improving the efficiency of embedded project development.