Privacy and Security for MapReduce Clouds with PASMAC

Lead PI

Co PI

Abstract

Despite the hype, only a few large enterprises or governmental organizations outsource their services to public clouds. On the one hand, cloud computing has been identified as one of the top ten business strategies as it offers many advantages such as greater flexibility and reduced costs. However, on the other hand, outsourcing and, therewith, relinquishing control of services implies many security and privacy problems. Cloud providers often place their data centers in foreign countries where security policies are difficult to enforce, cloud infrastructures are threatened by hackers, insiders, and even malicious customers trying to peek into or tamper with outsourced data. Consequently, the cloud cannot be trusted. As of today, lack of security and privacy guarantees are major adoption obstacles for both, large enterprises and governmental organizations.

PASMAC targets the design and evaluation of protocols for secure and privacy- preserving “data analysis” in an untrusted cloud. With PASMAC, the user can store and query data in the cloud, preserving privacy and integrity of outsourced data and queries. PASMAC specifically addresses a real-world cloud framework: Google’s prominent MapReduce paradigm. PASMAC will design and prototype new protocols based on highly parallelizable, efficient privacy-preserving techniques, such as efficient private information retrieval, encrypted Bloom filters, and additive homomorphic encryption.

For more information about PASMAC and its subprojects, visit the project page.

Funding

NSF

Related Publications