©2002 By Information Systems Audit and Control Association www.isaca.org F E AT U R E F E AT U R E Application Risk in a TCP/IP Environment By Robert M. Harrison I t often is said that the greatest risk to an organization comes from within its own walls. There are numerous cases where an employee intentionally causes harm to a company. Sometimes, a company can be exposed to risk simply by carelessness of its employees. This article addresses a common risk caused by carelessness in corporate networks. TCP/IP As a general rule, the flexibility of a networking technology is related inversely to the strength of its security capabilities. One networking technology, more than any other, was created with the specific goal of flexibility although it comes at the expense of virtually all inherent security. This technology is TCP/IP. TCP/IP is, by its very nature, a highly adaptable and fluid technology. However, these attributes do not come without cost in the form of increased security risks. In particular, specific applications using the TCP/IP technology can pose a major threat to every organization’s information assets without the proper monitoring tools in place. Scenario The network under review for this case study was a multistate Ethernet wide area network using several protocols with a variety of server, workstation and network operating systems. One aspect of the audit was to identify security weaknesses within the TCP/IP structure using common, easy-to-obtain tools used by hackers. We did not use any special hardware technology, and only used software that was freely available either on the Internet or on the company’s internal network. We worked with the network access that had been granted to us by the information security services department. This access was similar to any typical end user. Audit Process We began our audit by gathering data about the size and scope of the organization’s TCP/IP environment. This effort involved determining a range of TCP/IP addresses with which to work. We used our own address (obtained using the WINIPCFG command in Windows 98) as the starting point. Since we were operating within the organization’s firewall, it was easy to obtain other TCP/IP INFORMATION SYSTEMS CONTROL JOURNAL, VOLUME 6, 2002 addresses with a freeware port scanner. In less than one hour, we had identified all TCP/IP nodes at the company’s main office and all remote locations. Using this information, we noted that many of the host names assigned to individual PCs often either identified the end user or provided location information. This allowed us to zero in on the high-risk areas of the organization. However, despite this knowledge, we were unable to compromise the network operating systems or the operating systems of any of the PCs identified. The information security department had done an excellent job in securing the operating systems. However, we were able to spend a significant amount of time “hacking” without fear of being caught. There were no monitoring tools on the network that would have detected our actions and reported them to the appropriate personnel. At this point, we realized a different approach was required. Instead of attacking operating systems, this time, we would attempt to find weaknesses in applications commonly used at the company. Working within our own subnet, it took less than an hour before this effort yielded promising results. The organization was a heavy user of a popular remote desktop management application. This software allows a user at one PC to connect to another and assume control of that PC for remote maintenance. This control can be obtained through dial-up access or via a variety of networking protocols. In our case, the protocol exclusively in use was TCP/IP. By using the application itself, we were able to quickly identify all PCs on our subnet and were then able to connect and control the majority of these PCs. From there, we moved onto the wider network. Using our freeware port scanner, we quickly determined that the vast majority of PCs located outside of the company’s main office also had remote access software installed. Furthermore, none of these remote clients had password protection enabled, and all were configured to automatically startup when a PC was booted. Thus, turning on a PC made it vulnerable. The dangers of this situation were readily apparent. Among them were the ability to: • Obtain restricted company information • Obtain limited control over other employees’ e-mail accounts and messages • Change customer data 39 • Obtain limited control over several of the company’s core financial systems Since all applications were running in an unmonitored TCP/IP environment, our actions could not even be detected. We were not required to log on to any of the remote PCs to perform any of these activities. Robert M. Harrison is senior consultant, internal audit services, at J.H. Cohn LLP, an accounting and consulting firm. Harrison is based in Franklin Lakes, New Jersey, USA, and can be reached at rharrison@jhcohn.com. Lessons Learned Several factors contributed to the success of our penetration attacks. First, there was poor application management at the company. This was due to a lack of coordination between the network and technical support departments that allowed an application to be installed and configured in such a way that hundreds of PCs were vulnerable to attack. Second, the company did not have network detection and intrusion tools. At no time were the relevant people aware of our actions. Finally, although this particular application initially may have been set up to perform remote maintenance, it was not the ideal tool for this purpose. There are other, more secure ways to accomplish this task. Applications such as ZenWorks, Intel LANDesk and MS Systems Management Server can enable support personnel to perform maintenance on a remote PC from a central location without introducing easily identifiable and exploitable vulnerabilities to the corporate network. 40 INFORMATION SYSTEMS CONTROL JOURNAL, VOLUME 6, 2002