Towards Practical Differential Privacy for SQL Queries

By Noah M. Johnson, Joseph P. Near, Dawn Song. 2017

302 citations recorded by the source corpus.

Differential privacy promises to enable general data analytics while protecting individual privacy, but existing differential privacy mechanisms do not support the wide variety of features and databases used in real-world SQL-based analytics systems. This paper presents the first practical approach for differential privacy of SQL queries. Using 8.1 million real-world queries, we conduct an empirical study to determine the requirements for practical differential privacy, and discuss limitations of previous approaches in light of these requirements. To meet these requirements we propose elastic sensitivity, a novel method for approximating the local sensitivity of queries with general equijoins. We prove that elastic sensitivity is an upper bound on local sensitivity and can therefore be used to enforce differential privacy using any local sensitivity-based mechanism. We build FLEX, a practical end-to-end system to enforce differential privacy for SQL queries using elastic sensitivity. We demonstrate that FLEX is compatible with any existing database, can enforce differential privacy for real-world SQL queries, and incurs negligible (0.03%) performance overhead.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Prochlo: Strong Privacy for Analytics in the Crowd (2017, methodology): “By leveraging multiparty computations, they create virtual trusted-third-party platforms similar to the centralized PINQ or FLEX systems, which support differentially-private release of analysis results about user data [38, 52].”
Answering Multi-Dimensional Analytical Queries under Local Differential Privacy (2019, background): “, Apple [32], Google [16], Uber [23], and Microsoft [9].”
PriPeARL: A Framework for Privacy-Preserving Analytics and Reporting at LinkedIn (2018, methodology): “FLEX is a system designed to enforce di erential privacy for SQL queries using elastic sensitivity, and has been adopted by Uber for internal data analytics [20].”
Optimizing error of high-dimensional statistical queries under differential privacy (2018, background): “These techniques can answer queries using off-the-shelf systems (queries in SQL and data in relational form), and thus can be implemented efficiently [22,30].”
An Algorithmic Framework For Differentially Private Data Analysis on Trusted Processors (2018, background): “Because of this, all industrial deployments of DP by Apple, Google, and Microsoft, with the exception of Uber [32], have been set in the so called local model of differential privacy (LDP) [24, 19, 18].”
Proving differential privacy with shadow execution (2019, background): “Differential privacy is now considered a gold standard in privacy protections after recent high profile adoptions [1, 22, 27, 37].”
PROCHLO: Strong Privacy for Analytics in the Crowd (2017, methodology): “The analyzer, which has the most complete view of the reported data, may choose from a wide variety of differentially private mechanisms and data processing systems [27,31,38, 53, 63].”
Differentially Private SQL with Bounded User Contribution (2019, methodology): “For our experiments with PINQ and Flex, we also set sensitivity to our previously determined ∆uQ1 listed in Table 3.”
PrivateSQL: A Differentially Private SQL Query Engine (2019, background): “In comparison, a system that uses the state-of-the-art Flex [24] incurs > 100% error for over 65% of the queries; i.”
Hypothesis Testing Interpretations and Renyi Differential Privacy (2019, background): “Over the last years, differential privacy has found use in the private sector [Kenthapadi et al., 2019] by companies such as Google [Erlingsson et al., 2014, Papernot et al., 2018], Apple [team at Apple, 2017], and Uber [Johnson et al., 2018], and in the public sector by agencies such as the U.S. Ce”