Explore a conference talk on SECRECY, a groundbreaking system for privacy-preserving collaborative analytics in cloud environments. Discover how this innovative approach allows multiple data holders to contribute to joint analyses while maintaining data privacy, even from cloud providers. Learn about the system's use of cryptographically secure Multi-Party Computation (MPC) to ensure no information leakage and provide provable security guarantees. Delve into SECRECY's novel optimization techniques, including co-designing multiple layers of the system stack and exposing MPC costs to the query engine. Understand how physical and logical optimizations dramatically improve performance, reducing computation, communication, and space requirements during query execution. Gain insights from multi-cloud experiments demonstrating SECRECY's ability to improve query performance by over 1000x compared to existing approaches, enabling complex analytics on millions of data records with modest resource usage.
Overview
Syllabus
NSDI '23 - SECRECY: Secure collaborative analytics in untrusted clouds
Taught by
USENIX