Intelligent reconfigurable surfaces (IRSs) can enhance the communication performance by smartly reconfiguring the wireless communication environment. In particular, the phase shifts can be configured to change the small-scale fading properties of the wireless channels, and the location and azimuth of the RIS panel can be optimally deployed to compensate the large-scale path loss. The former has been extensively studied in the literature while the azimuth deployment problem remains to be explored. In this paper, the azimuth deployment is studied for the downlink multiple-input single-output IRS-aided multi-user system. Our goal is to maximize the weighted sum rate (WSR) of users by jointly optimizing the active beamforming at the BS, the passive beamforming and the azimuth of the IRS, subject to the minimum rate constraint of each user and the total power constraint. An alternating optimization (AO) algorithm is proposed to obtain the optimal solutions. We also derive the closed-form expressions for the optimal solutions for the single user case, which provides tractable results and useful insights for the multiple user case in practice. Simulation results show that the WSR can be improved by 20.36% when IRS is deployed with the optimal azimuth. Moreover, it is necessary to focus the azimuth of the IRS to the users with higher QoS requirements.