With the privacy issues drawing more and more concerns, privacy protection techniques based on Computational Private Information Retrieval (CPIR) allow a user to retrieve data from a service provider without revealing the users query information. For largescale applications, there exists a gap between privacy protection techniques and its feasibility. For the problem that the CPIR algorithm needs long computing time so as not to be suitable for largescale data privacy protection, this paper proposes a CPIR nearest neighbor privacy protection algorithm (HPCPIRV) based on Spark and Huffman code. The HPCPIRV algorithm partitions the spatial data into Voronoi diagrams according to the points of interest in the data preprocessing stage, and then utilizes the Huffman code to compress the candidate data in order to reduce bit computation operation. Spark parallel framework is used for query grid parallel computing in the server side. The experimental results show that the computational cost of HPCPIRV algorithm is about 30% lower than that of PCPIRV algorithm on the server side, the computational cost of client is about 10% lower, and the communication cost is about 40% lower.