There were no overwhelming surprises in the SANS Institute's much-quoted annual list of Top 20 Internet security risks -- unless you count the fact that there were just 18 threats on the list this time around.
The list for 2007, which was released earlier this week, marks the first time that the SANS Institute has deviated from its practice of categorizing threats by operating system and software platforms. Instead, the focus for 2007 was on a "functional division" of vulnerabilities, according to Rohit Dhamankar, leader of the SANS Top 20 project and a senior manager of research at Austin-based security vendor TippingPoint.
As a result, this year's list of risks was organized under six broad categories that included client-side vulnerabilities, server-side vulnerabilities, security policy, and personnel and application abuse. The risks listed under each category were largely the same as last year, but many of them have grown far more severe.
Among the most significant takeaways from this year's list:
Client-side threats are on the rise. The attention that software vendors and their users have paid to addressing server-side risks has resulted in a much more hardened server environment in comparison with a few years ago. Malicious attackers are finding fewer opportunities to take advantage of flaws on the server side, so they are increasingly focusing their attention on the client side.
"What's new here is the sheer number of client-side vulnerabilities that are being discovered these days," said Amol Sarwate, manager of vulnerability labs at Qualys Inc. and a longtime contributor to the list. "Flaws in universally used client software such as word processors, spread sheets, PDF viewers, media players, Web browsers and e-mail clients present a clear and growing danger for companies and consumers alike. ... It's high time administrators stop focusing on server threats alone and focus instead on flaws in products such as these because they are so widely used and cut across operating systems and browser platforms."
Complicating the issue, added Sarwate, is the fact that many gullible users often unwittingly help attackers take advantage of such flaws by falling victim to phishing and other social engineering attacks.
"People who are on the desktop are becoming the biggest threat, especially if they have unfiltered access to the Internet," Dhamankar said. "There are literally hundreds of attack vectors on the client side, and the people who are coding exploits are coming out with clever mechanisms for evading detection."
Web application threats are bigger than ever. As with client-side threats, Web software risks are generally well understood and have been a security focus for some time now. But once again it is the sheer number of flaws that are being discovered in Web-based applications -- including custom-developed ones and others such as content management system wikis, portals and bulletin boards -- that makes it a hot-button issue, according to Dhamankar. Every week, he noted, literally hundreds of new Web application vulnerabilities are being discovered, with many of them being actively exploited -- in fact, a good 50% of the 4,396 vulnerabilities report by SANS between November 2006 and October 2007 were Web application flaws.
The most common Web app attacks are SQL injection, cross-site scripting and cross-site request forgery flaws, according to the SANS list. The number of attempted attacks per day against such flaws range from hundreds of thousands to even millions for some of the large Web hosting farms.
Mobile, voice-over-IP (VoIP), IPv6 and zero-day threats were not as big as they were expected to be. Certainly, each of these are big issues and risks that are not to be taken lightly, according to Dhamankar and Sarwate. But for various reasons, the risks associated with each of these areas did not materialize quite as dramatically as had been predicted. In the case of VoIP and IPv6, the fact that these technologies still haven't been broadly adopted likely made them somewhat less of an attractive target to attackers, Sarwate said. Similarly, several zero-day attacks, which are attacks against flaws for which no patch yet exists, were recorded in 2007. But that number was lower than in 2006 -- opposite to what many analysts had predicted would happen.