No silver bullet

17.04.2006
First, the bad news. Then, the worse news. We are years away from having a single security architecture to protect company information. That's because every area in IT has different technical hurdles to cross before security can be assured. And in each segment today, we are a long way from satisfactory protection.

Evalubase Research Inc. released a survey in February that covered five technology areas: data management, hardware and operating systems, communications and networking, application development, and industry applications. The research firm asked IT professionals to rank those technologies for performance, usability, functionality, compatibility, maintainability and security. You won't be shocked to learn that security ranked at the bottom for all except hardware and operating systems, and communications and networking, where it was ranked next to last.

Nick Caffarra, president of Evalubase, tells me that maybe, maybe, in five or more years there could be an integrated cross-technology security approach from one vendor capable of protecting your information. But he doesn't sound optimistic.

Little wonder that he isn't bullish on a single security approach, because here's the worse news. It comes from Seth Hallem, CEO of Coverity Inc. His company scans source code for defects, most of which lead to security holes. (The Department of Homeland Security and Stanford University chose Coverity to analyze open-source tools for defects.) Hallem points to research that proves it's mathematically impossible to eliminate defects from source code. Mathematically impossible.

So, that's the news. Your company's information isn't secure today, and it won't ever be.

It's All Relative

Of course, security is relative. Last month, the folks at Coverity released some data for defect scans on 31 open-source projects. The average defect rate for 1,000 lines of source code was 0.42. Not bad. A programmer would, on average, crank out 2,200 lines of code for each flub. But if that rate were constant against, say, the 30 million lines of Red Hat Linux 7.1, you'd have 12,600 lines with problems. If it held steady against the 213 million lines of source in Debian 3.1, you'd find 89,460 potential defects.

This isn't to say that Debian is less secure as a server operating system than Red Hat. Or vice versa. But it does point to the kind of information you can use to lower the risk your information faces. That is, you can use tools to quantify your risk and then decide when, where and whether to use a technology.

Common Sense

You can also use common-sense strategies to protect your company's information. Would your data be inherently more secure if more end users had Macintoshes? Despite news in February that the first (benign) virus for the Mac was discovered, the answer would have to be yes. That's because viruses and worms written for one system wouldn't be propagated by the other. In other words, a mix of operating systems is a good defensive strategy.

Do all end users really need fat clients -- Windows or Macs? Would some be able to get their work done more securely on thin clients? Of course. A mix of thin clients, Macs and Windows, as well as different server systems, is an ideal defense against many of today's vulnerabilities. A side benefit is challenging the skills of hackers who will try to penetrate your defenses with primarily Windows-specific knowledge.

Some single-platform advocates argue that the IT costs of running multiple operating systems make it problematic to run a mixed environment. Maybe so. But these people have a short-term view of cost. The costs of a security breach are far greater. The University of Maryland estimates that when a public company suffers a single security breach, its market capitalization drops 5 percent. Would you want to tell the board of directors not to worry because the company saved some of that shareholder value in IT support costs through your single-platform strategy?

Business = Risk

Every business faces risk the moment it opens its doors. IT's job is to keep the risk to information at a minimum. Hoping for one solution -- the security silver bullet -- isn't realistic. The one-way approach has proved to be a security liability when implemented as a uniform platform strategy.

Given how valuable information is to a company, Evalubase's Caffarra says it might be time to put corporate data on a company's balance sheet as an asset. If that happened, maybe the board would insist that the very best tools and methodologies be applied to decrease the risk to that information. And that the very best strategy isn't to put all your eggs in one basket.

Mark Hall is a Computerworld editor at large. Contact him at mark_hall@computerworld.com.