Collaborative Research: SaTC: CORE: Medium: Mixed Distribution Models for Encrypted Data Stores

Many organizations today capture and store an immense amount of data in cloud storage systems. Unfortunately, simply storing the data in an encrypted form is insufficient to ensure security: an adversary can launch powerful attacks that use data access patterns and data lengths to learn damaging information about the data. Existing countermeasures against these attacks remain impractical at the scale of modern storage systems due to large bandwidth and/or storage overheads.

This state of affairs leave organizations with two equally expensive and painful options: (1) encrypted but insecure systems; or (2) secure systems but with orders-of-magnitude higher operational costs.

This project breaks the above impasse by exploring a fundamentally new mixed distribution model that has the potential to both enable secure and high-performance data stores, as well as to provide new insights about the less-explored tradeoff between adversarial strength and performance limits of cloud storage systems. Thus, this project enables the design of cloud storage systems that resist important adversaries in practice, getting both formal security analyses and efficiency.

This project has broad ramifications for how practitioners navigate the security-performance tradeoff while building cloud storage systems; for many scenarios, the project enables breaking the (unfortunate) choice between security and performance offered by existing solutions, paving the path for finally addressing security flaws that have typically required orders-of-magnitude higher operational costs. The project outcomes are open-source research artifacts, and broader community impact via educational and technology transfer activities.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.