Cloud storage providers build a distributed storage system by utilizing cloud resources located in data centers. The interactions among servers in a DHT (Distributed Hash Table)-based cloud storage system depend on the routing process, and its execution logic is more complicated. Hence, how to allocate resources to not only guarantee service performance (e.g., data availability, response delay), but also help service providers to reduce cost became a challenge. To address this challenge, this paper presents a novel resource provisioning model for cloud storage systems. The model utilizes queuing network for analysis of both service performance level and cost calculation. Then the problem is defined as a cost optimization with performance constrains, and a novel algorithm is proposed. Furthermore, we implemented a DHT-based storage system on top of an infrastructure platform built with OpenStack. Based on real-world traces collected from our system, we show that our model could effectively guarantee the target data availability and response delay with lower cost.