Dan Geer wrote an article for SC Magazine on The enterprise information protection paradigm, discussing the fundamental disconnect between the derived value of data and the investment to protect information. He asks the important question: If we reap ever increasing returns on information, where is the investment to protect the data? Dan has an eloquent take on a long-standing viewpoint in the security community that Enterprise Information Protection (EIP) is a custodial responsibility of corporations, as it is core to generation of revenue and thus the company’s value.
Dan’s point that we don’t pay enough attention (and spend enough money and time) on data security is inarguable – we lose a lot of data, and it costs. His argument that we should concentrate on (unification of) existing technologies (such as encryption, audit, NAC, and DLP), however, is flawed – we already have lots of this technology, so more of the same is not the answer.
Part of our problem is that in the real world, inherent security is only part of the answer. We also have external support, such as police who arrest bank robbers – it’s not entirely up to the bank to stop bank robbers. In the computer security world – for various reasons – legal enforcement is highly problematic and much less aggressive than for physical crimes like robbery.
I don’t have a problem with Dan’s reasoning on this issue. His argument for the motivation to secure information is sound. I do, however, take issue with a couple of the examples he uses to bridge his reasoning from one point to the next.
First, Dan states, “We have spent centuries learning about securing the physical world, plus a few years learning about securing the digital world. What we know to be common to both is this: That which cannot be tolerated must be prevented.” He puts that in very absolute terms, and I do not believe it is true in either the physical or electronic realms. For example, our society absolutely does not tolerate bank robberies. However, preventative measures are miniscule. The banks are open for business and pretty much anyone can walk in the door. Rather than prevent a robbery, we collect information from witnesses, security cameras, and other forensic information – to find, catch, and punish bank robbers. We hope that the threat of the penalty will deter most potential robbers, and sound police work will allow us to catch up with the remainder who are daring enough to commit these crimes.
While criminals are very good at extracting real value from virtual objects, law enforcement has done a crappy job at investigating, punishing, and (indirectly) deterring crimes in and around data theft. These two crucial factors are absent in electronic crimes in comparison to physical crimes. It’s not that we can’t – it’s that we don’t.
This is not to undermine Dan’s basic point – that enterprises which derive value from data are not protecting themselves sufficiently, and contributorily negligent. But stating that “The EIP mechanism – an unblinking eye focused on information – has to live where the data lives.” and “EIP unifies data leakage prevention (DLP), network access control (NAC), encryption policy and enforcement, audit and forensics,” argues that network and infrastructure security are the answer. As Gunnar Peterson has so astutely pointed out many times, while the majority of IT spending is in data management applications, our security spending is predominately in and around the network. That means the investments made today are to secure data at rest and data in motion, rather than data in use. Talking about EIP as an embodiment of NAC & DLP and encryption policy reinforces the same suspect security investment choices we have been making for some time. We know how to effectively secure data “at that point where data-at-rest becomes data-in-motion”. The problem is we suck ” … at the point of use where data is truly put at risk …” – that’s not network or infrastructure, but rather in applications.
A basic problem with data security is that we do not punish crimes at anywhere near the same rate as we do physical crimes. There is no (or almost no) deterrence, because examples of capturing and punishing crimes are missing. Further, investment in data security is typically misguided. I understand how this happens – protecting data in use is much harder than encrypting TCP/IP or disk drives – but where we invest is a critical part of the issue. I don’t want this to come across as disagreement with Dan’s underlying premise, but I do want to stress that we need to make more than one evolutionary shift.