It is never impossible for a hacker to break into a computer system, only improbable. Computer hackers routinely break into corporate, military, online banking, and other networked environments. Even in 2007, as I am writing this second edition of Network Security Assessment, I still perform incident response work in these sectors. As systems generally become more secure, the methods used by these attackers are becoming more advanced, involving intricate repositioning, social engineering, physical compromise (stealing disks from servers or installing rogue wireless access points), and use of specific zero-day exploits to attack peripheral software components such as antivirus or backup solutions that are widely deployed internally within corporate networks.

By the same token, you would expect professional security consultants to be testing for these types of issues. In the vast majority of cases they are not. I know this because at Matta we run a program called Sentinel, which involves testing security assessment vendors for companies in the financial services sector. The Sentinel platform contains a number of vulnerable systems, and vendors are scored based on the vulnerabilities they identify and report. Since 2004, Matta has processed nearly 30 global penetration testing vendors using Sentinel. In a recent test involving 10 testing providers, we found the following:

• Two vendors failed to scan all 65536 TCP ports

• Five vendors failed to report the publicly accessible MySQLservice root password of “password”

• Seven vendors failed to report the easily exploitable, high-risk SSLPCT overflow (MS04-011)

A number of vendors have tested the Sentinel platform on more than one occasion. It is clear that there is a lack of adherence to a strict testing methodology, and test results (in particular, the final report presented to the customer) vary wildly, depending on the consultant involved.