Abstract: In order to improve the rationality of distributed cloud storage system construction, a minimum redundant distributed cloud storage system construction method under heterogeneous environment is proposed.Firstly, an analytical framework for computing data availability in heterogeneous environments is developed.Because the computation of data availability is very complicated when the storage nodes are growing, a Monte Carlo method is proposed to estimate the real values with less computational cost.Secondly, since it is difficult to compute the optimal redundancy allocation for each host, a method based on particle swarm optimization(PSO)is proposed.Finally, a simple iterative algorithm is provided to determine the minimum redundancy required to ensure data availability for different storage applications.Experimental results show that the proposed method can reduce data redundancy and storage cost.