The feedback of massive stars and clusters, such as radiation, wind, and supernovae, accounts for most of the energy budget in galaxies. However, our understanding of massive stars and cluster formation is still poor. Statistics studies of various evolutionary stages of massive star formation are crucial to reach a comprehensive understanding of massive star and cluster formation. In order to investigate the formation of massive stars and their associated cluster, we carried out a systematic program with the interferometers (e.g., ALMA, SMA, JVLA) and single dish telescopes (e.g., IRAM-30m, SMT, CSO) toward different evolutionary stages of massive star-forming regions. Our studies yield promising clues to the formation of massive stars and protoclusters. For instance, we find that the nonthermal motions are predominantly subsonic and transonic, and gas filament widths are narrower than the previously proposed ‘quasi-universal’ 0.1 pc filament width. In this talk, I will present a detailed of our recent results using different observations from 10 pc scales to a few 0.01 pc scales, including turbulence, filament, prestellar/protostelar core,  molecular outflows, accretion, fragmentation, etc.