Extensions
- anon 0.5.0
- Data Anonymization for Postgres
Documentation
README
Contents
Anonymization & Data Masking for PostgreSQL
postgresql_anonymizer
is an extension to mask or replace
personally identifiable information (PII) or commercially sensitive data from
a PostgreSQL database.
The projet is aiming toward a declarative approach of anonymization. This means we're trying to extend PostgreSQL Data Definition Language (DDL) in order to specify the anonymization strategy inside the table definition itself.
Once the maskings rules are defined, you can access the anonymized data in 3
different ways :
- Anonymous Dumps : Simply export the masked data into an SQL file
- In-Place Anonymization : Remove the PII according to the rules
- Dynamic Masking : Hide PII only for the masked users
In addition, various Masking Functions are available : randomization, faking, partial scrambling, shufflin, noise or even your own custom function !
Read the Concepts section for more details and NEWS.md for information about the latest version.
Declaring The Masking Rules
The main idea of this extension is to offer anonymization by design.
The data masking rules should be written by the people who develop the application because they have the best knowledge of how the data model works. Therefore masking rules must be implemented directly inside the database schema.
This allows to mask the data directly inside the PostgreSQL instance without using an external tool and thus limiting the exposure and the risks of data leak.
The data masking rules are declared simply by using security labels :
```sql =# CREATE EXTENSION IF NOT EXISTS anon CASCADE;
=# SELECT anon.load();
=# CREATE TABLE player( id SERIAL, name TEXT, points INT);
=# SECURITY LABEL FOR anon ON COLUMN player.name -# IS 'MASKED WITH FUNCTION anon.fake_last_name()';
=# SECURITY LABEL FOR anon ON COLUMN player.id -# IS 'MASKED WITH VALUE NULL'; ```
In-Place Anonymization
You can permanetly remove the PII from a database with anon.anymize_database()
.
This will destroy the original data. Use with care.
```sql =# SELECT * FROM customer; id | full_name | birth | employer | zipcode | fk_shop -----+------------------+------------+---------------+---------+--------- 911 | Chuck Norris | 1940-03-10 | Texas Rangers | 75001 | 12 112 | David Hasselhoff | 1952-07-17 | Baywatch | 90001 | 423
=# CREATE EXTENSION IF NOT EXISTS anon CASCADE; =# SELECT anon.load();
=# SECURITY LABEL FOR anon ON COLUMN customer.full_name -# IS 'MASKED WITH FUNCTION anon.fake_first_name() || '' '' || anon.fake_last_name()';
=# SECURITY LABEL FOR anon ON COLUMN customer.birth
-# IS 'MASKED WITH FUNCTION anon.random_date_between(''01/01/1920''::DATE,now())';
=# SECURITY LABEL FOR anon ON COLUMN customer.employer -# IS 'MASKED WITH FUNCTION anon.fake_company()';
=# SECURITY LABEL FOR anon ON COLUMN customer.zipcode -# IS 'MASKED WITH FUNCTION anon.random_zip()';
=# SELECT anon.anonymize_database();
=# SELECT * FROM customer; id | full_name | birth | employer | zipcode | fk_shop -----+-------------------+------------+------------------+---------+--------- 911 | michel Duffus | 1970-03-24 | Body Expressions | 63824 | 12 112 | andromache Tulip | 1921-03-24 | Dot Darcy | 38199 | 423
```
You can also use anonymize_table()
and anonymize_column()
to remove data
from a subset of the database.
Dynamic Masking
You can hide the PII from a role by declaring it as a "MASKED". Other roles will still access the original data.
Example:
sql
=# SELECT * FROM people;
id | fistname | lastname | phone
----+----------+----------+------------
T1 | Sarah | Conor | 0609110911
(1 row)
Step 1 : Activate the dynamic masking engine
sql
=# CREATE EXTENSION IF NOT EXISTS anon CASCADE;
=# SELECT anon.start_dynamic_masking();
Step 2 : Declare a masked user
sql
=# CREATE ROLE skynet LOGIN;
=# SECURITY LABEL FOR anon ON ROLE skynet IS 'MASKED';
Step 3 : Declare the masking rules
```sql =# SECURITY LABEL FOR anon ON COLUMN people.lastname -# IS 'MASKED WITH FUNCTION anon.fake_last_name()';
=# SECURITY LABEL FOR anon ON COLUMN people.phone -# IS 'MASKED WITH FUNCTION anon.partial(phone,2,$$**$$,2)'; ```
Step 4 : Connect with the masked user
sql
=# \! psql peopledb -U skynet -c 'SELECT * FROM people;'
id | fistname | lastname | phone
----+----------+-----------+------------
T1 | Sarah | Stranahan | 06******11
(1 row)
Anonymous Dumps
Due to the core design of this extension, you cannot use pg_dump
with a masked
user. If you want to export the entire database with the anonymized data, you
must use the anon.dump()
function :
console
$ psql [...] -qtA -c 'SELECT anon.dump()' your_dabatase > dump.sql
NB: The -qtA
flags are required.
Warning
This is projet is at an early stage of development and should used carefully.
We need your feedback and ideas ! Let us know what you think of this tool,how it fits your needs and what features are missing.
You can either open an issue or send a message at contact@dalibo.com.
Requirements
This extension is officially supported on PostgreSQL 9.6 and later. It should also work on PostgreSQL 9.5 with a bit of hacking. See the Developement Notes for more details.
It requires two extensions :
tsm_system_rows which is delivered by the
postgresql-contrib
package of the main linux distributionsddlx a very cool DDL extrator
Install
- Install the extension on the server with :
console
sudo pgxn install ddlx
sudo pgxn install postgresql_anonymizer
- Add 'anon' in the
shared_preload_libraries
parameter of youpostgresql.conf
file. For example:
shared_preload_libraries = 'pg_stat_statements, anon'
- Restart your instance.
You can also read the INSTALL section for detailed instructions or if you want to deploy it on Amazon RDS or some other DBAAS service.
Limitations
- The dynamic masking system only works with one schema (by default
public
). When you start the masking engine withstart_dynamic_masking()
, you can specify the schema that will be masked withSELECT start_dynamic_masking('sales');
. However in-place anonymization withanon.anonymize()
and anonymous export withanon.dump()
will work fine with multiple schemas.
Performance
So far, we've done very few performance tests. Depending on the size of your data set and number of columns your need to anonymize, you might end up with a very slow process.
Here's some ideas :
Sampling
If your need to anonymize data for testing purpose, chances are that a smaller subset of your database will be enough. In that case, you can easily speed up the anonymization by downsizing the volume of data. There are mulitple way to extract a sample of database :
Materialized Views
Dynamic masking is not always required ! In some cases, it is more efficient to build Materialized Views instead.
For instance:
sql
CREATE MATERIALIZED VIEW masked_customer AS
SELECT
id,
anon.random_last_name() AS name,
anon.random_date_between('01/01/1920'::DATE,now()) AS birth,
fk_last_order,
store_id
FROM customer;