Risk-analysis tools provide the big security picture

All computer security defense ultimately comes down to managing risk. Security admins implement various defenses, each of which should have its own cost/benefit analysis. The cost of the defense should not outweigh the estimated damage of the attack or exploit. For example, if buying anti-malware software for a 100 PCs costs $3,900 per year, but cleaning up the damage from a malware attack would cost only $2,000 per year, implementing the anti-malware software wouldn’t make sense.

On the other hand, perhaps it would be reasonable to deploy anti-malware only on an external gateway server for the cost of $1,500 per year while being prepared to absorb the potential $250 cost of cleaning up a PC that gets infected by a USB storage stick. Computer security is rarely a binary decision — it’s shades of gray analyzed across the spectrum using risk analysis.

[ Learn how a rough economy creates opportunities for better IT security. | Tune in to the InfoWorld Security Central channel for the latest IT security news and reviews. ]

Lately, while working on several forthcoming product reviews, I’ve come across some software products that make risk assessment easier to visualize and to present to management. Probably the best example is Bit9’s Parity whitelisting application-control program. Parity comes with a service that automatically recognizes 5 billion different files (according to Bit9) based upon their hash signature and other file attributes. Parity can recognize every relatively recent file made by Microsoft, Adobe, and Macromedia, as well as waves of lesser-known programs, including common hacking tools, such as Nmap, Pwdump, Cain & Abel, and so on.

Here’s how it work: Parity first compares computers to a baseline “gold standard” image established by the administrator. As users install new software (or malware exploits), Parity is able to calculate a “drift” score for the computer. By accumulating all the drift scores for all the computers on the network, Parity can calculate an aggregate drift score.

Even better, Parity also calculates a file’s threat risk. For example, if it were to find a new text editor, it would give the program a low threat rating, because it’s unlikely to be involved with an exploit. But if Parity discovers an unauthorized peer-to-peer or popularly exploited program, that program’s risk rating would be higher. And if it happens upon a hacking tool — say pwdump, used to dump Windows password hashes — that program would have a high risk rating, as would any identified malware program.

Parity calculates a risk rating for every file, then compiles those scores to calculate a risk rating for the computer. From there, Parity pools the risk ratings of all of the managed computers to compute a rating for the entire organization. The beauty of this methodology is that IT security can report the risk rating to management and compare it month to month. If IT security is successful in lowering the risk rating, management will easily be able to see the decrease in security risk. I love it!

Another whitelisting program in the same review, SignaCert, uses a similar approach, but the risk ratings are calculated and input by the administrator: same principle, but more work. ArcSight’s Logger has a similar risk-ranking ability if used in conjunction with the company’s Enterprise Security Manager product. I’m sure there are lots of other products that do the same thing. Readers, please send me your risk-ranking product candidates.

As a frequent security product Grinch, I have to say that these risk-ranking products please me. They are an end-goal product in that they take the actual security defense and allow the IT security administrator to easily calculate a metric that senior management cares about, and they make it easy to demonstrate improved compliance and decreased risk over time. Who couldn’t love such a product?

Risk-ranking products are far from perfect, however. First, no two companies apply the same risk value to a given file. My company may be very risk-averse to installing BearShare, for example, whereas other companies may not care. Fortunately, most of these types of products allow administrators to customize risk rankings to fit their needs.

Second, a low-ranked risk product may actually end up being high risk — for example, Adobe Reader and Microsoft Office, to name two.

Third, these products can measure managed machines only. Unmanaged machines are present in any network and are usually higher risk than the average computer, almost by definition.

Fourth, unrecognized files might not have any risk ranking at all, though they can usually be assigned a common risk value by default. Should unrecognized files be defaulted to high or low risk? You might think “high risk” by default, but then again, you’d be surprised by how many files are unrecognized. If you choose the former setting as your default, you could be imputing higher risk than your environment really has.

Fifth, rating existing files is only one aspect of security risk. Patches, end-user behavior, anti-malware scanners, and all the other deployed security defenses impacts a computer’s overall risk.

I recognize that risk ranking by file existence is imperfect, but with so few security metrics that really mean something in the computer world, I’m willing to celebrate what we have when we have it.

Speaking of better imperfect technologies, take a humor break with this clip.