6. GDPR Module

This module supports the data warehouse in meeting GDPR requirements by anonymising personal data in the historised data. It is designed to anonymise data of individuals that has been deleted or obscured in the source systems, e.g. after an erasure request or after the end of a retention period.

Configuring the anonymisation requires a domain expert familiar with the tables, their meaning, and their purpose. All configurations are recorded historically, so that they can be presented in audit inspections.

Note

The GDPR module overwrites personal data in place. Physical deletion of whole rows after a retention period is handled by the delete logic (erase groups).

6.1. Process

The GDPR component resides in the schema REDGPR, which contains the metadata repository for configuring and controlling the anonymisation: which PSA tables are anonymised and which rules (trigger view and GDPR view) are used.

The rules are customer-specific. The Application Manager creates two kinds of views, by default in the schema REDGPR_CUS:

Trigger view

Delivers the business keys (RED_BSK) of all subjects that are to be anonymised, e.g. customers whose relationship ended 10 years ago. The column RED_BSK is mandatory.

GDPR view

Delivers, for one PSA table, all rows to be anonymised, identified by their primary key (RED_PK), together with the anonymised values of the affected columns, e.g. a birth date set to 1 January of the decade. Besides RED_PK the view must only contain the anonymised columns; these must exist in the PSA table, and no other RED_* columns are allowed.

Important

The GDPR view itself must restrict its result to the business keys delivered by the trigger view. The trigger view only controls whether an anonymisation run is executed for the PSA table at all (i.e. whether it is not empty).

GDPR run: trigger view, GDPR view and configuration checks feed the anonymisation run, which overwrites columns in the PSA table and writes a run protocol Customer-specific rules Trigger view RED_BSK of affected subjects GDPR view RED_PK and anonymised values Configuration checks mandatory and existing columns Anonymisation run weekly job not empty no errors PSA table columns overwritten in place Run protocol rows, status, errors
GDPR anonymisation run

The anonymisation run is executed as a job:

  1. The configuration of all active mappings is checked (e.g. mandatory columns, columns existing in the PSA table). Mappings with errors are skipped and the errors are logged.

  2. For each active mapping whose trigger view is not empty, the anonymised values of the GDPR view overwrite the corresponding columns of the PSA rows. The keys and time slices of the rows are not changed; each changed row is marked with the ID of the run.

  3. Each run is recorded with the number of updated rows and its status.

The jobs are inactive after the installation and are scheduled weekly (Sunday) once activated. Trigger views and mappings are inactive by default as well.

Warning

The framework overwrites the anonymised columns. The original values can only be restored from a database backup. Anonymisation may also cause existing logical constraints to show inconsistencies. Check the expected results of the views carefully before activating a trigger or mapping.

Benefits of the anonymisation of historical data in RED:

  • Every configuration is auditable.

  • Flexible and configurable for any source.

  • Configuration checks before each run.

  • The results can be checked in advance by querying the views before the trigger and the mapping are activated.