The combination of phase-sensitive optical time-domain reflectometry (F-OTDR) and deep learning algorithms is a popular research topic in recent years. To train the classification models established by these algorithms, a huge amount of data is required. Though F-OTDR can monitor the applied perturbations along the sensing fiber continuously, the huge volume of data puts a heavy burden on storage devices. In this paper, we propose a lossy data compression method based on quantization technology to solve the data storage problem in heterodyne F-OTDR. Experimental results show that the vibration waveform can be successfully restored from the reconstructed signal. A compression ratio of 16 is achieved by quantifying a 128-MB data to 31.25 MB with 293.2-ms time consumption. With the proposed quantization method, it is expected to be able to store more sensing data without making modifications to the hardware.