This is important because focusing on an insider or outsider is a red herring to database security efforts. Insider theft is a specific threat model -- but just one to be considered.
Access controls provide the front line of defense, but access and authorization can be obtained without credentials. So if we reduce permissions to a minimum and keep the settings perfectly in line with established policies, there are dozens of ways to directly or indirectly obtain access and authorization. The most common is to compromise a service that has credentials, and then the attacker reprograms the service to do the dirty work. Sure, hackers guess passwords and sniff them off the network, but in many data breaches, access controls are bypassed entirely.
But access controls don't verify usage. Phil Lieberman of Lieberman Software captured this position in the Dark Reading piece:
"Problems like using commonly known shared passwords, never changing sensitive passwords, and allowing their employees to have too much access for too long to sensitive data with no accountability is the rule rather than the exception,"
You would think that better access controls and better administration that keep settings up to date was the best way to address the threat. But accountability is the real issue. Once you have access, you can perform any function that your authorization profile allows. Which is exactly what Mr. Falciani did at HSBC. The problem is he was not caught until he tried to sell the information to someone outside the company.
The important point Mr. Lieberman makes is the lack of accountability. Taking a page from accounting practices, proper separation of duties coupled with auditing are the most basic elements of fraud detection, and absent from most database security operations. If there is no way to perform validation for activity, there is no way to detect fraud electronically, and you are reliant on external systems (the French government in this case). External parties (customers, partners, peers) have been the common element in detecting most of the major data breaches, further evidence internal controls are absent or inadequate.
If I have guessed the password for an admin account, and I queries the customer database, how can you tell if I am an insider or an outsider? Can you determine if the activity is part of my normal job function, or am I stealing data? The common modes of access into the database, queries and extraction methods are leveraged by attackers and legitimate user alike. Sometimes there is no way to tell the difference. In other cases, context and behavior offer clues to detect and even stop data theft. When the request is coming from outside the company, at odd times a day, from an unknown application or simply exhibit irrational query patterns we have a very good indication of misuse.
The insider threat will always be a problem, regardless of how good your access control scheme is, because you have to provide employees credentials to do their job, and you have to trust them at some point. If you are worried about data theft, database monitoring and auditing are essential measures for fraud detection and security. These technologies close the gap left by access control systems for many different types of threats, not just employee theft.
Adrian Lane is an analyst/CTO with Securosis LLC, an independent security consulting practice. Special to Dark Reading.