PIRMAP: Efficient Private Information Retrieval for MapReduce

Private Information Retrieval (PIR) allows a user to retrieve bits from a database while hiding the user’s access pattern. However, the practicality of PIR in a real-world cloud computing setting has recently been questioned. In such a setting, PIR’s enormous computation and communication overhead is expected to outweigh the cost saving advantages of cloud computing. In this paper, we first examine existing PIR protocols, analyzing their efficiency and practicality in realistic cloud settings. We identify shortcomings and, subsequently, present an efficient protocol (PIRMAP) that is particularly suited to MapReduce, a widely used cloud computing paradigm. PIRMAP focuses especially on the retrieval of large files from the cloud, where it achieves good communication complexity with query times significantly faster than previous schemes. To achieve this, PIRMAP enhance related work to allow for optimal parallel computation during the “Map” phase of MapReduce, and homomorphic aggregation in the “Reduce” phase. To improve computational cost, we also employ a new, faster “somewhat homomorphic” encryption, making our scheme practical for databases of useful size while still keeping communication costs low. PIRMAP has been implemented and tested in Amazon’s public cloud with database sizes of up to 1 TByte. Our evaluation shows that non-trivial PIR such as PIRMAP can be more than one order of magnitude cheaper and faster than trivial PIR in the real-world.