Abstract:  For the massive data of GNSS observation, the method of conventional centralized data processing is facing the problem of low computational efficiency, distributed to process the baseline by using the technology of Hadoop platform, which can solve this problem effectively. By dividing the subnet, calling the GAMIT software to process the baseline in parallel, the result of 63 IGS stations shows that the acceleration ratio of the three compute nodes is 6.67, and the accuracy of the coordinate points is in the order of millimeters.