Rome Laboratory, located in the rolling hills of Upstate New York, is the United States Air Force's premier command-and-control research facility. It conducts projects on everything from artificial intelligence to radar guidance to target detection. In April 1994, IT personnel discovered that hackers had broken into the Rome network. For at least three days, the cybercriminals had unrestricted access to the system and were able to copy and download classified files. The monetary cost of the intrusion was significant -- at least $500,000 -- but the security risk was much greater. "We have only the intruders to thank for the fact that no lasting damage occurred," stated the official Air Force report. "Had they decided, as a skilled attacker most certainly will, to bring down the network immediately after the initial intrusion, we would have been powerless to stop them." 

Unfortunately for the Department of Defense (DOD), the Rome Laboratory intrusion was not an isolated incident. (See "Hacking the DOD" on the next page.) The U.S. General Accounting Office reports that hackers attempt to break into DOD computers at least 250,000 times a year. Too often, they succeed. But the lessons the DOD has learned from combating the hackers can be applied to risk management and IT security everywhere.

HOW THE HACKERS DO IT

Securing a computer network from hackers may be compared to protecting a house against burglars. The files are the valuables the thief wants. The internet ports are the doors and windows through which he may enter. The user is the homeowner. And in recent years, it is the users that have generally been a hacker's favorite targets. 

In the Rome Laboratory incident, hackers sent emails to lab personnel that contained Trojan horse files in a technique known as "spear phishing," a directed strategy that has proven far more dangerous than indiscriminate, mass-mail "phishing." When employees opened the files, they unknowingly loaded malware onto their computers and gave the hackers access. It was as if they had handed a house key to a burglar. Having gotten their foot in the door, the hackers set up fake identities within the Rome network and used these phony profiles to launch attacks on other facilities in locales as far flung as California, South Korea and Latvia. The hackers may have been young, energetic, imaginative geeks who viewed network security as a puzzle to solve and enjoyed pitting themselves against the collective brains of the DOD, but they could just as well have been working for a foreign country that paid them to steal research or sabotage the country's military network.  

The 2007 hack into the Office of the Secretary of Defense (OSD) was in some ways even more serious than the Rome Laboratory intrusion. Undetected for two months, malicious code spread throughout the systems, allowing the hackers to steal data from the computers. Repairing the damage and updating the system cost $4 million. Once again, the hackers used email to exploit software vulnerability, sending fake messages to employees to steal their user IDs and passwords. The messages were carefully crafted for the recipients, containing familiar names in order to convince them that the email requests were legitimate. 

COUNTERMEASURES

Social-engineering attacks such as those launched against the OSD are surprisingly effective -- so much so that some organizations use simulated attacks to train their users. In 2004, more than 500 West Point cadets received an email from Col. Robert Melville alerting them to a problem with their grade reports and instructing them to click on an embedded link to verify their grades. More than 80% of the cadets clicked on the link. But there was no Col. Robert Melville and no problems with the grades. 

The message was a test sent by Aaron Ferguson, a computer security expert who taught at West Point. Training drills like this are controversial. Some security personnel believe playing "gotcha" with workers is going too far and erodes trust in the organization. They favor using software to authenticate emails. Ideally, however, both user training and technical screening should be part of the security equation.

Accordingly, the OSD took a multipronged approach to recover from the 2007 intrusion. The compromised network was shut down for three weeks, during which time a new process for checking temporary IDs and passwords was installed. Other safeguards applied included tighter requirements for using common access cards to verify identities and digital signatures to ensure that information comes from a valid source. 

As part of the new emphasis on cybersecurity, federal agencies were directed by the White House Office of Management and Budget to reduce their external connections from 1,300 to no more than 50. Since connections to the outside world are like doors and windows in a house, fewer connections mean fewer avenues for a hacker to sneak through. 

Mischel Kwon, former chief IT security specialist for the Justice Department, waxed enthusiastic over this "Trusted Internet Connections" initiative. "This is an absolutely great, great program," she told a group of IT professionals at a 2007 meeting of the Association for Federal Information Resources Management. 

Other security experts were more reserved. Chris Wysopal, chief technology officer at application security vendor Veracode sees portal security as only part of the solution. "There is no easy way to filter all the bad data coming in over internet ports," he said. "It is too complex. There are web application firewalls that can filter out simple attacks and there are now web gateways that can block users from accessing some malicious sites. But they are not a substitute for building secure web applications and keeping workstations patched. Software needs to have an individual security assessment before it is deployed." 

Can a network be 100% protected? Can a house be completely burglar-proof? Perhaps not. But if we follow the lessons learned by the DOD and take an integrated approach to security, considering all vulnerabilities -- software, users and ports -- we will go a long way towards reducing the risk of an expensive and perhaps devastating intrusion. 

----------

Dr. William H. Stevenson, III is a freelance writer and chemical consultant who has worked with a variety of government contractors for more than 20 years.