Share on Facebook Tweet on Twitter Share on LinkedIn Share by email
Database Security and Compliance
Database Security and Compliance

Much of the sensitive data in any organization is stored in a database system. There is a natural need to manage the security of the sensitive information. Our goal in this project is to develop tools to manage database security.

There are two broad facets of the problem we address.

Fine Grained Authorization

We are working on developing the database system infrastructure for supporting fine grained authorizations. The idea is that there is an authorization policy in place that determines how data can be accessed by users. Unlike the SQL standard that supports only coarse grained authorization our goal is to permit fine-grained access to data. The second and related challenge is efficient query processing. Since the policy is applicable for all data access, every query run in the system has to be made compliant (by potentially rewriting it). Clearly, there is a tradeoff between the expressivity of the policy language and the query performance. On the one hand, if the policy is expressed through arbitrary user defined code, then enforcing it may impose a high overhead on query processing. Our challenge is to design an authorization infrastructure that simultaneously allows us to express rich policies without adversely impacting query performance.

Data Auditing

The second facet of data compliance we address is auditing. Database systems have support to monitor various operations performed against the database server. This audit trail is used offline to perform ad-hoc analysis of potential security breaches and threats. In fact, commercial tools are available that help analyze events such as login failures and schema changes. However, the support for data auditing is rather limited. The goal of data auditing is to ask whether any of the queries and updates that were run against the system potentially breached information about sensitive data. The key challenges in data auditing are: 1) semantics: what does it mean for a query to potentially breach sensitive information and 2) performance: we are often analyzing audit trails spanning long periods of time (weeks or even months) during which a large number of queries and updates can be run. Thus it becomes essential to make the auditing as efficient as possible.

Publications
People
Surajit Chaudhuri
Surajit Chaudhuri

Raghav Kaushik
Raghav Kaushik

Ravi Ramamurthy
Ravi Ramamurthy