With the development of software defined networks (SDN), the measurement approaches have got more and more attention. Most of the existing researches focused on globally measurement to reduce the cost and improve the accuracy. Nevertheless, fewer research employed multiple measurement points to reduce the measurement cost. Thus, in this paper, we address the problem of placing multiple measurement points to measure the network parameters, called MMPPR. We first present the model of the measurement cost, and then formulate the MMPPR as a 0-1 nonlinear programming problem and transform it into a 0-1 linear programming problem. At last, we propose an efficient algorithm for MMPPR based on Lagrange relaxation. Experimental result shows that our algorithm performs better compared with the previous algorithm.