Abstract: Aiming at the insufficient in the researches of parallel programming models for big data processing, this paper combines BSP model with virtual computing environment, and presents a hybrid distributed-memory and shared-memory parallel programming model BSPCloud which can take the advantages of BSP model's predictable performance, easily programming and deadlock avoidance. We have implemented a proof-of-concept BSPCloud parallel programming library in java. The performance predictability and speedup are evaluated in virtual computing environment. the experiment results display good speedup and predictability.