Surface roughness is an important feature of the product quality. In this paper, the function of ensuring workpiece processing quality and reducing the detection time of surface roughness is realized by remote monitoring surface roughness. The surface roughness is indirectly monitored by monitoring the spindle power, workpiece vibration, and cutting parameters. Using the method of support vector regression (SVR), four prediction models based on the X-, Y-, and Z-directions' vibration, spindle power, and cutting parameters are established through the comprehensive comparison of the five evaluation criteria. The prediction models based on the X-direction's vibration and spindle power are the best. A remote monitor system for surface roughness is established using the data acquisition unit, cloud servers, and client side to monitor the operation of the machine remotely. And when the surface roughness is abnormal, the remote alarm will be realized to facilitate timely inspection. Moreover, users can trace the defective parts by looking up the history. In this way we can find the source of abnormal processing. [ABSTRACT FROM AUTHOR]