Protecting National Infrastructure
By Edward G. Amoroso
Copyright © 2011 Elsevier Inc.
All right reserved.
Chapter One INTRODUCTION
Somewhere in his writings—and I regret having forgotten where—John Von Neumann draws attention to what seemed to him a contrast. He remarked that for simple mechanisms it is often easier to describe how they work than what they do, while for more complicated mechanisms it was usually the other way round. Edsger W. Dijkstra
National infrastructure refers to the complex, underlying delivery and support systems for all large-scale services considered absolutely essential to a nation. These services include emergency response, law enforcement databases, supervisory control and data acquisition (SCADA) systems, power control networks, military support services, consumer entertainment systems, financial applications, and mobile telecommunications. Some national services are provided directly by government, but most are provided by commercial groups such as Internet service providers, airlines, and banks. In addition, certain services considered essential to one nation might include infrastructure support that is controlled by organizations from another nation. This global interdependency is consistent with the trends referred to collectively by Thomas Friedman as a "flat world."
National infrastructure, especially in the United States, has always been vulnerable to malicious physical attacks such as equipment tampering, cable cuts, facility bombing, and asset theft. The events of September 11, 2001, for example, are the most prominent and recent instance of a massive physical attack directed at national infrastructure. During the past couple of decades, however, vast portions of national infrastructure have become reliant on software, computers, and networks. This reliance typically includes remote access, often over the Internet, to the systems that control national services. Adversaries thus can initiate cyber attacks on infrastructure using worms, viruses, leaks, and the like. These attacks indirectly target national infrastructure through their associated automated controls systems (see Figure 1.1).
A seemingly obvious approach to dealing with this national cyber threat would involve the use of well-known computer security techniques. After all, computer security has matured substantially in the past couple of decades, and considerable expertise now exists on how to protect software, computers, and networks. In such a national scheme, safeguards such as firewalls, intrusion detection systems, antivirus software, passwords, scanners, audit trails, and encryption would be directly embedded into infrastructure, just as they are currently in small-scale environments. These national security systems would be connected to a centralized threat management system, and incident response would follow a familiar sort of enterprise process. Furthermore, to ensure security policy compliance, one would expect the usual programs of end-user awareness, security training, and third-party audit to be directed toward the people building and operating national infrastructure. Virtually every national infrastructure protection initiative proposed to date has followed this seemingly straightforward path.
While well-known computer security techniques will certainly be useful for national infrastructure, most practical experience to date suggests that this conventional approach will not be sufficient. A primary reason is the size, scale, and scope inherent in complex national infrastructure. For example, where an enterprise might involve manageably sized assets, national infrastructure will require unusually powerful computing support with the ability to handle enormous volumes of data. Such volumes will easily exceed the storage and processing capacity of typical enterprise security tools such as a commercial threat management system. Unfortunately, this incompatibility conflicts with current initiatives in government and industry to reduce costs through the use of common commercial off-the-shelf products.
In addition, whereas enterprise systems can rely on manual intervention by a local expert during a security disaster, large-scale national infrastructure generally requires a carefully orchestrated response by teams of security experts using predetermined processes. These teams of experts will often work in different groups, organizations, or even countries. In the worst cases, they will cooperate only if forced by government, often sharing just the minimum amount of information to avoid legal consequences. An additional problem is that the complexity associated with national infrastructure leads to the bizarre situation where response teams often have partial or incorrect understanding about how the underlying systems work. For these reasons, seemingly convenient attempts to apply existing small-scale security processes to large-scale infrastructure attacks will ultimately fail (see Figure 1.2).
As a result, a brand-new type of national infrastructure protection methodology is required—one that combines the best elements of existing computer and network security techniques with the unique and difficult challenges associated with complex, large-scale national services. This book offers just such a protection methodology for national infrastructure. It is based on a quarter century of practical experience designing, building, and operating cyber security systems for government, commercial, and consumer infrastructure. It is represented as a series of protection principles that can be applied to new or existing systems. Because of the unique needs of national infrastructure, especially its massive size, scale, and scope, some aspects of the methodology will be unfamiliar to the computer security community. In fact, certain elements of the approach, such as our favorable view of "security through obscurity," might appear in direct conflict with conventional views of how computers and networks should be protected.
National Cyber Threats, Vulnerabilities, and Attacks
Conventional computer security is based on the oft-repeated taxonomy of security threats which includes confidentiality, integrity, availability, and theft. In the broadest sense, all four diverse threat types will have applicability in national infrastructure. For example, protections are required equally to deal with sensitive information leaks (confidentiality), worms affecting the operation of some critical application (integrity), botnets knocking out an important system (availability), or citizens having their identities compromised (theft). Certainly, the availability threat to national services must be viewed as particularly important, given the nature of the threat and its relation to national assets. One should thus expect particular attention to availability threats to national infrastructure. Nevertheless, it makes sense to acknowledge that all four types of security threats in the conventional taxonomy of computer security must be addressed in any national infrastructure protection methodology.
Vulnerabilities are more difficult to associate with any taxonomy. Obviously, national infrastructure must address well-known problems such as improperly configured equipment, poorly designed local area networks, unpatched system software, exploitable bugs in application code, and locally disgruntled employees. The problem is that the most fundamental vulnerability in national infrastructure involves the staggering complexity inherent in the underlying systems. This complexity is so pervasive that many times security incidents uncover aspects of computing functionality that were previously unknown to anyone, including sometimes the system designers. Furthermore, in certain cases, the optimal security solution involves simplifying and cleaning up poorly conceived infrastructure. This is bad news, because most large organizations are inept at simplifying much of anything.
The best one can do for a comprehensive view of the vulnerabilities associated with national infrastructure is to address their relative exploitation points. This can be done with an abstract national infrastructure cyber security model that includes three types of malicious adversaries: external adversary (hackers on the Internet), internal adversary (trusted insiders), and supplier adversary (vendors and partners). Using this model, three exploitation points emerge for national infrastructure: remote access (Internet and telework), system administration and normal usage (management and use of software, computers, and networks), and supply chain (procurement and outsourcing) (see Figure 1.3).
These three exploitation points and three types of adversaries can be associated with a variety of possible motivations for initiating either a full or test attack on national infrastructure.
Each of the three exploitation points might be utilized in a cyber attack on national infrastructure. For example, a supplier might use a poorly designed supply chain to insert Trojan horse code into a software component that controls some national asset, or a hacker on the Internet might take advantage of some unprotected Internet access point to break into a vulnerable service. Similarly, an insider might use trusted access for either system administration or normal system usage to create an attack. The potential also exists for an external adversary to gain valuable insider access through patient, measured means, such as gaining employment in an infrastructure-supporting organization and then becoming trusted through a long process of work performance. In each case, the possibility exists that a limited type of engagement might be performed as part of a planned test or exercise. This seems especially likely if the attack is country or terrorist sponsored, because it is consistent with past practice.
At each exploitation point, the vulnerability being used might be a well-known problem previously reported in an authoritative public advisory, or it could be a proprietary issue kept hidden by a local organization. It is entirely appropriate for a recognized authority to make a detailed public vulnerability advisory if the benefits of notifying the good guys outweigh the risks of alerting the bad guys. This cost–benefit result usually occurs when many organizations can directly benefit from the information and can thus take immediate action. When the reported vulnerability is unique and isolated, however, then reporting the details might be irresponsible, especially if the notification process does not enable a more timely fix. This is a key issue, because many government authorities continue to consider new rules for mandatory reporting. If the information being demanded is not properly protected, then the reporting process might result in more harm than good.
Perhaps the most insidious type of attack that exists today is the botnet. In short, a botnet involves remote control of a collection of compromised end-user machines, usually broadband-connected PCs. The controlled end-user machines, which are referred to as bots, are programmed to attack some target that is designated by the botnet controller. The attack is tough to stop because end-user machines are typically administered in an ineffective manner. Furthermore, once the attack begins, it occurs from sources potentially scattered across geographic, political, and service provider boundaries. Perhaps worse, bots are programmed to take commands from multiple controller systems, so any attempts to destroy a given controller result in the bots simply homing to another one.
Excerpted from Cyber Attacks by Edward G. Amoroso Copyright © 2011 by Elsevier Inc.. Excerpted by permission of Butterworth-Heinemann. All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.