Data security and data privacy are important, no one disputes that. We all want to keep private things private and to keep our data secure. And yet, data needs to be shared, to enable insights, to help organizations observe patterns and have those “ah-ha” moments. None of us want the extreme where, in an effort to keep data secure, there is no access to data of any form within your organization, and the result is no business insights or analytics. With GDPR going into effect, you’ve likely been rethinking what security controls you have in place.
Here at Citus Data we collaborate with SaaS businesses and larger enterprises alike, generally to consult on Postgres data models and how to best scale out their database. (Our Citus extension to Postgres enables you to scale out Postgres horizontally. The benefit: performance.) In working with teams, one common thing we’ve seen companies do is to restrict who can see which bits of Personally Identifiable Information (PII) within your database. There are a number of approaches, including heavyweight ETL processes that mask PII bits. An ETL process tends to introduce a certain amount of latency from the time data is in your system until the time it can be analyzed.
Fortunately, Postgres provides a few primitives that can be used directly within your database to hide PII, while still enabling sophisticated analytics and exploration of data in real time.
Here we’ll look at using Postgres schemas and views to provide access to data while keeping PII safe and hidden. Keep reading