Comment by cgio

That should be captured in security, right? I.e. if someone has access to data the assumption is they are able to read it. Security audit can be done without the grain of select statements but rather the grain of changes to provisioning. Fundamentally, you have an issue even if someone did not actually read what they should not but could. Select audit can be useful for performance tuning but I would not classify it as “audit”, rather monitoring.