Statement of Dr. Steve Cross
Director, Software Engineering Institute
February 23, 2000
Introduction
Mr. Chairman and Members of the Joint Economic Committee:
My name is Steve Cross. I am the director of the Software Engineering Institute (SEI) at Carnegie Mellon University. The SEI is the home of the CERT® Coordination Center (CERT/CC). Thank you for the opportunity to testify on the issue of computer security and the US economy. Today I will give you some background on the CERT/CC, describe the trends we have observed while responding to computer security incidents on the Internet, discuss near-term steps that I believe can be taken to address today’s problems, and consider what the future holds.
The CERT® Coordination Center (CERT/CC) is part of the Survivable Systems Initiative of the Software Engineering Institute, a federally funded research and development center operated by Carnegie Mellon University. The CERT/CC was established more than eleven years ago, after an Internet "worm" stopped 10% of the computers connected to the Internet. This program—the first Internet security incident to make headline news—was the wake-up call for network security. In response, the CERT/CC was established at the SEI. Its charter was to work with the Internet community to respond to computer security events, raise awareness of computer security issues, and prevent security breaches. While continuing to respond to incidents, the CERT/CC provides training, investigates tools and techniques that enable typical users and administrators to effectively protect systems from damage caused by intruders, conducts research leading to increased security of the Internet, and serves as a model to others establishing incident response teams. The CERT/CC is now recognized by both government and industry as a neutral, authoritative source of information assurance information and expertise. More details about our work are attached to the end of this testimony (see Meet the CERT Coordination Center).
In the first full year of operation, 1989, The CERT/CC responded to 132 computer security incidents. In 1999, the staff responded to more than 8,000 incidents. In total, the CERT/CC staff has handled well over 24,000 incidents and analyzed more than 1,500 computer vulnerabilities. This testimony is based on that first-hand experience.
Economy’s Dependence on the Internet
Doing business over the Internet is quick, easy, and inexpensive—compelling reasons for companies to turn to electronic commerce. Last year an estimated 23 million shoppers spent more than $11 billion. Business-to-business transactions were even larger—$671 billion was exchanged in 1998, $92 billion in Internet transactions and $579 billion using EDI. Those figures are expected to rise significantly every year for the foreseeable future, with a predicted business-to-business growth rate in e-commerce of 33% each year. Experts predict that the US will reach $1.3 trillion in business-to-business transactions by 2003, with one-fourth of all US business-to-business purchases done online.
Much of the growth will come from the continued expansion of the Internet. An additional 16 million hosts connected to the Internet between July 1999 and January 2000, bringing the estimated total to 72.4 million. Of that number, nearly 25 million hosts are at commercial (.com) sites.
Threats to Commerce and Their Impact
When companies rely on online business transactions, they must protect their information from exposure or tampering, protect the privacy of their customers’ information, and protect themselves against network "downtime" and denial-of-service attacks. They need to earn and maintain their customers’ trust, and they need stay open for business.
The current state of Internet security is cause for concern. Evidence can be found in the results of several recent surveys. For example, of 500 organizations contacted for the 1999 CSI/FBI survey, 62% reported computer security breaches within the last year. Of those, 38% reported from 2 to 5 incidents, and 26% reported 10 or more incidents. Those reporting their Internet connection as a frequent point of attack rose for a third straight year, from 37% in 1996 to 57% in 1999. Twenty-six percent reported theft of proprietary information (an increase of 8% over 1998).
The survey reported that 96% of the 521 survey respondents have Web sites, and 30% of them provide e-commerce services. Of the Web and e-commerce sites, 20% had detected unauthorized access or misuse of the sites within the last 12 months. Even more worrisome is the fact that 33% answered that they didn’t know if their site had been compromised.
Financial losses from network security breaches are costly. They regularly exceed $100 million a year. The top causes are loss of proprietary information ($42.5 million) and financial fraud ($39.7 million). Yankee Group estimates that the denial-of-service attacks on web sites earlier this month could cost the companies more than $1.2 billion. On the days of the attacks, losses exceeded $100 million. Service disruptions ranged from two and three-quarter hours to five hours. Looking to the future, the FBI estimates that computer crime will cost companies $10 billion next year.
A survey published by Information Security magazine (July 1999) found that companies conducting business online are 57% more likely to suffer leaks of proprietary information than companies that are not on the Web. And the rate at which intruders are breaching corporate networks has nearly doubled in the last year. Overall, companies suffered an average loss of $256,000 to security breaches last year. Of the 745 organizations surveyed, 91 quantified their financial losses for a total of $23.3 million.
The White House Office of Science and Technology estimates an annual cost of $100 million for US losses of proprietary information. The American Society for Information Science (ASIS) estimates that the losses may exceed $250 billion. Additionally, costly damage can be done by changing price lists, discount rates, or interest rates. Untold costs in loss of business can occur when customers lose faith in a company’s ability to protect information such as credit card numbers, names, addresses, credit information, and the like. In August 1999, PC Magazine reported on the "real security hot buttons," commenting that "All of the recent online shopping studies confirm that privacy and security are the most important issues to consumers by far. You wouldn’t trust a bank that didn't lock its doors at night; why would you trust a Web site that was similarly open?"
After a successful computer system intrusion, it can be very difficult or impossible to determine precisely what subtle damage, if any, was left by the intruder. Loss of confidence can result even if an intruder leaves no damage because the site cannot prove none was left. Particularly serious for business are denial-of-service attacks and the exposure of sensitive information. Once an overt denial-of-service attack has been resolved and the service returned, customers generally regain trust in the service they receive. But exposure of sensitive information makes a businesses highly susceptible to a loss-of-confidence crisis.
Here are just a few examples of security breaches that have been reported in the press. In addition to these examples, the CERT/CC handles reports of breaches at e-commerce sites daily.
It is obvious from these examples and the ongoing activity of the CERT Coordination Center that there is much work to be done to secure our electronic networks adequately to meet the needs of the expanding e-commerce marketplace. The next section discusses why such problems exist, and the final sections suggests actions to address the growing security issues relating to commerce on the Internet.
Vulnerability of the Internet and World Wide Web
Vulnerabilities associated with the Internet put both business and individual users at risk. Security measures that were appropriate for mainframe computers and small, well-defined networks inside an organization, are not effective for the Internet, a complex, dynamic world of interconnected networks with no clear boundaries and no central control. Because the Internet was not originally designed with security in mind, it is difficult to ensure the integrity, availability, and privacy of information. The Internet was designed to be "open," with distributed control and mutual trust among users. As a result, control is in the hands of users, not in the hands of the provider; and use cannot be administered by a central authority. Furthermore, security issues are not well understood and are rarely given high priority by software developers, vendors, network managers, or consumers.
In addition, because the Internet is digital, not physical, it has no geographic location and no well-defined boundaries. Traditional physical "rules" are difficult or impossible to apply. Instead, new knowledge and a new point of view are required to understand the workings and the vulnerabilities of the Internet.
Another factor is the approach typically taken by the intruder community. There is (loosely) organized development in the intruder community, with only a few months elapsing between "beta" software and active use in attacks. Moreover, intruders take an open-source approach to development. One can draw parallels with open system development: there are many developers and a large, reusable code base.
Intruder tools are becoming increasingly sophisticated and also becoming increasingly user friendly and widely available. For the first time, intruders are developing techniques to harness the power of hundreds of thousands of vulnerable systems on the Internet. Using what are called distributed-system attack tools, intruders can involve a large number of sites simultaneously, focusing all of them to attack one or more victim hosts or networks. The sophisticated developers of intruder programs package their tools into user-friendly forms and make them widely available. As a result, even unsophisticated intruders can use them.
The current state of Internet security is the result of many additional factors, such as the ones listed below. A change in any one of these can change the level of Internet security and survivability.
Attack Strategies Illustrating Internet Vulnerabilities
Some attacks are intended to harass a site and deny it the ability to transact business on the Internet. Other attacks enable intruders to gain privileged access to a system so that it effectively belongs to them. With their unauthorized privileges, they can, for example, use the system as a launch platform for attacks on other sites or as one node in an attack using distributed-system intruder tools. Still other attacks are designed to reveal sensitive information, such as passwords or trade secrets. We describe sample attack strategies below. Our descriptions are neither theoretical nor abstract; rather, they present, at a high level, actual attacks reported to the CERT Coordination Center regularly.
Use of Distributed System Intruder Tools
Distributed systems based on the client/server model have become increasingly common. In recent months, the CERT/CC has seen an increase in the development and use of distributed sniffers, scanners, and denial-of-service tools. Attacks using these tools can involve a large number of sites simultaneously and be focused to attack one or more victim hosts or networks.
Damaged systems include those used in the attack as well as the targeted victim. For the victim, the impact can be extensive. For example, in a denial-of-service attack using distributed technology, the attacked system observes simultaneous attacks from all the nodes at once– flooding the network normally used to communicate and trace the attacks and preventing any legitimate traffic from traversing the network.
There are indications that the processes for discovering vulnerable sites, compromising them, installing daemons (programs used in the attack), and concealing the intrusion are largely automated, with each step being performed in "batch" mode against many machines in one "session." Daemons have been discovered on a variety of operating systems with varying levels of security and system management.
Planning and coordination before an attack are critical to ensuring adequate response when the attack is in progress. Since the attack methodology is complex and there is no single-point solution or "silver bullet," resolution and restoration of systems may be time-consuming. The bottom line is that an organization’s systems may be subject at any time to distributed attacks that are extremely difficult to trace or defend against.
Although an organization may be able to "harden" its own systems to help prevent implantation of the daemon portion of a distributed attack tool, there is essentially nothing a site can do with currently available technology to prevent becoming a victim of, for example, a coordinated network flood. The impact upon the site and its operations is dictated by the (in)security of other sites and the ability of a remote attacker to implant the tools and subsequently to control and direct multiple systems worldwide to launch an attack. The result may be reduced or absent network connectivity for extended periods of time, possibly days or even weeks depending upon the number of sites attacking and the number of possible attack networks that could be activated in parallel or sequentially.
Coordinated attacks across national boundaries have been observed. The tools and attacks demonstrate that a network that optimizes its technology for speed and reliability at the expense of security may experience neither speed nor reliability, as intruders abuse the network or deny its services. The intruder technology is evolving, and future tools may be more difficult to defeat.
Web-Specific Attacks
The CERT/CC has published a number of security alerts about Web-related technology (attached), the most recent of which concerned cross-site scripting. In this attack, what users receive from a web site may not be what the operators of that site meant to send. Attackers put malicious code into otherwise legitimate HTML code so that if users click on a specially designed link, they may receive bad data, unwanted pictures, and programs (malicious scripts) to compromise users’ data. The attackers can capture passwords and other information the users believe is protected, and they may be able to view protected parts of the users’ network, such as an intranet.
The problem is not with web browsers themselves but with how dynamic web pages are constructed (dynamic means they are constructed "on the fly" in response to user input) and how data entering and leaving web sites is validated. "Validate" means ensuring no "unintended" characters are sent back to the client. The attack is possible because web browsers have the capability to interpret scripts embedded in web pages downloaded from a web server. Web browsers are usually installed with the capability turned on by default. The user can unknowingly download the script when visiting a seemingly safe site and completing an interactive form or querying a database; following untrustworthy links in web pages, email, or newsgroups; and viewing dynamically generated web pages. The malicious script then runs on the user’s browser. Although attackers have been able to inject malicious code for a long time, the cross-site scripting attack is significant because users can acquire malicious code from legitimate, typically trustworthy sites; avoiding questionable sites is no longer adequate protection.
Most solutions require action from a broad community of web page developers and web site administrators. They must ensure that their web pages are encoded in a way that neutralizes malicious code, apply patches developed by vendors, and filter all data that enters and leaves web servers. In the meantime, users can gain some protection by turning off certain features of their web browsers, limiting the functionality they may be accustomed to having. They cannot fully protect themselves nor easily identify an attack.
The World Wide Web is a young and still immature technology. Until it is "hardened" for safe, effective use, it will continue to be vulnerable to security compromises – and online companies and customers alike will continue to be concerned about the integrity and privacy of information that must be exchanged if they are to do business on the Web.
SYN Attacks: Denial of Service
A SYN attack is an attack against a computer that provides service to customers over the Internet. SYN refers to the type of message (Synchronize) that is used between computers when a network connection is being made. In this attack, the enemy runs a program from a remote location (anywhere in the world) that jams the service on the victim computer. This is known as a denial-of-service attack because the effect of the attack is to prevent the service-providing computer from providing the service. The attack might prevent one site from being able to exchange data with other sites or prevent the site from using the Internet at all. Increasingly, companies are depending on Internet services for day-to-day business, from email to advertising to online product delivery. Some companies’ business is entirely dependent on the Internet.
SYN attacks have been used successfully against a wide variety of targets, but they have the greatest impact against Internet service providers, or ISPs, which provide Internet connection services to government, businesses, and individuals. A SYN attack against an ISP usually results in disruption of Internet service to all the service provider’s customers.
This type of attack is very difficult to prevent because it exploits a design flaw in the basic technology used for Internet communication today. Experts are currently working on techniques to reduce the problem somewhat, but preventing these attacks from occurring in the future will require a change in the way Internet communications are accomplished by the computers using the Internet. This is likely to take several years.
IP Spoofing: Masquerading
In an attack known as IP spoofing, attackers run a software tool that creates Internet messages that appear to come, not from the intruder’s actual location, but from a computer trusted by the victim. IP, which stands for Internet Protocol, refers to the unique address of a computer. When two computers trust each other, they allow access to sensitive information that is not generally available to other computer systems. The attacker takes advantage of this trust by masquerading as the trusted computer to gain access to sensitive areas or take control of the victim computer by running "privileged" programs. Information that has been compromised through IP spoofing includes credit card information from a major Internet service provider and exploitation scripts that a legitimate user had on hand for a security analysis.
Unfortunately, there are many computer programs and services that rely on other computers to "speak the truth" about their address and have no other mechanism for disallowing access to sensitive information and programs. The CERT Coordination Center has received many reports of attacks in which intruders (even novice intruders) used this technique to gain access to computer systems with the help of publicly available IP spoofing computer programs.
Sniffers: Violating Privacy and Confidentiality
For most users of computer networks, including the Internet, the expectation is that once a message is sent to another computer or address, it will be protected in much the same way letters are protected in the U.S. Postal Service. Unfortunately, this is not the case on the Internet today. The messages are treated more like postcards sent by a very fast, efficient pony express. Information (such as electronic mail, requests for connections to other systems, and other data) is sent from one computer to another in a form easily readable by anyone connected to a part of the network joining the two systems together. For Internet data, these messages are routed through the networks at many locations, any one of which could choose to read and store the data as it goes by. The CERT/CC has handled many incidents in which an intruder ran a program known as a sniffer at a junction point of the Internet.
The sniffer program records many kinds of information for later retrieval by the intruder. Of specific interest to most intruders is the user name and password information used in requests to connect to remote computers. With this information, an intruder can attack a computer on the Internet using the name and password of an unsuspecting Internet user. Intruders have captured hundreds of thousands of these user name/password combinations from major companies, governments sites, and universities all over the world.
To prevent attacks of this type, encryption technology must be used for both the access to other computers around the Internet (cryptographic authentication) and the transmission of data across the Internet (data encryption).
Attractiveness of the Internet to Attackers
Compared with other critical infrastructures, the Internet seems to be a virtual breeding ground for attackers. Although some attacks seem playful (for example, students experimenting with the capability of the network) and some are clearly malicious, all have the potential of doing damage. Unfortunately, Internet attacks in general, and denial-of-service attacks in particular, remain easy to accomplish, hard to trace, and a low risk to the attacker.
Internet Attacks Are Easy
Internet users place unwarranted trust in the network. It is common for sites to be unaware of the amount of trust they actually place in the infrastructure of the Internet and its protocols. Unfortunately, the Internet was originally designed for robustness from attacks or events that were external to the Internet infrastructure, that is, physical attacks against the underlying physical wires and computers that make up the system. The Internet was not designed to withstand internal attacks—attacks by people who are part of the network; and now that the Internet has grown to encompass so many sites, millions of users are effectively inside.
The Internet is primarily based on protocols (rules and conventions) for sharing electronically stored information, and a break-in is not physical as it would be in the case of a power plant, for example. It is one thing to be able to break into a power plant, cause some damage, then escape. But if a power plant were like the Internet, intruders would be able to stay inside the plant undetected for weeks. They would come out at night to wander through the plant, dodging a few guards and browsing through offices for sensitive information. They would hitch a ride on the plant’s vehicles to gain access to other plants, cloning themselves if they wished to be in both places at once.
Internet attacks are easy in other ways. It is true that some attacks require technical knowledge—the equivalent to that of a college graduate who majored in computer science—but many successful attacks are carried out by technically unsophisticated intruders. As mentioned earlier, technically competent intruders duplicate and share their programs and information at little cost, thus enabling naive "wanna-be" intruders to do the same damage as the experts.
Internet Attacks Are Difficult to Trace
As discussed in the IP spoofing example, attackers can lie about their identity and location on the network. Information on the Internet is transmitted in packets, each containing information about the origin and destination. Again, a packet can be compared to a postcard—senders provide their return address, but they can lie about it. Most of the Internet is designed merely to forward packets one step closer to their destination with no attempt to make a record of their source. There is not even a "postmark" to indicate generally where a packet originated. It requires close cooperation among sites and up-to-date equipment to trace malicious packets during an attack.
Moreover, the Internet is designed to allow packets to flow easily across geographical, administrative, and political boundaries. Consequently, cooperation in tracing a single attack may involve multiple organizations and jurisdictions, most of which are not directly affected by the attack and may have little incentive to invest time and resources in the effort.
This means that it is easy for an adversary to use a foreign site to launch attacks at US systems. The attacker enjoys the added safety of the need for international cooperation in order to trace the attack, compounded by impediments to legal investigations. We have seen US-based attacks on US sites gain this safety by first breaking into one or more non-US sites before coming back to attack the desired target in the US.
Internet Attacks Are Low Risk
Failed attempts to break into physical infrastructures involve a number of federal offenses; such events have a long history of successful prosecutions. This is not the case for Internet intrusions. Because attacks against the Internet typically do not require the attacker to be physically present at the site of the attack, the risk of being identified is reduced. In addition, it is not always clear when certain events should be cause for alarm. For example, what appear to be probes and unsuccessful attacks may actually be the legitimate activity of network managers checking the security of their systems. Even in cases where organizations monitor their systems for illegitimate activity, which occurs in only a small minority of Internet-connected sites, real break-ins often go undetected because it is difficult to identify illegitimate activity. In the case of cross-site scripting, web users trigger malicious code without even knowing they have done so, and web sites can unknowingly pass the code along. Finally, because intruders cross multiple geographical and legal domains, an additional cloud is thrown over the legal issues involved in pursuing and prosecuting them.
Recommended Solutions
The problem is serious and complex, and a combination of approaches must be used to reduce the risks associated with the ever-increasing dependence on the Internet and the possibility of a sustained attack on it. Effective solutions require multi-disciplinary and cross-domain cooperation that includes information sharing and joint development of comprehensive solutions, as well as support for a long-term research agenda.
Support an established center for collecting, analyzing, and disseminating information assurance information.
The nature of threats to the Internet is changing rapidly and will continue to do so for the foreseeable future. The combination of rapidly changing technology, rapidly expanding use, and the continuously new and often unimagined uses of the Internet creates a volatile situation in which the nature of threats and vulnerabilities is difficult to assess and even more difficult to predict.
To help ensure the survivability of the Internet, and the information infrastructure as a whole, it is essential to continuously monitor and analyze cybersecurity threats and vulnerabilities and identify trends in intrusion activity. The organization doing this should collect, analyze, and report on quantity, trends, and character of cybersecurity incidents. To obtain the required information, the organization must be well trusted throughout the community. Given the universal concerns about privacy and confidentiality and the inherently voluntary nature of reporting, the collection organization should be neither government nor commercial. Nor can it be responsible for public policy, investigation, enforcement, or other activities perceived as conflicting. Organizations that have suffered attacks are often unwilling to discuss their problems for fear of loss of confidence by their customers.
The CERT/CC is establishing an analysis center to expand its work of collecting and analyzing information assurance data. The goals are to identify trends and to develop detection and mitigation strategies that provide high-leverage solutions to information assurance problems, including countermeasures for new vulnerabilities and emerging threats. It takes advantage of the information dissemination channels already in place at the CERT/CC.
The CERT Analysis Center extends current incident response capabilities by developing and transitioning protective measures and mitigation strategies to defend against advanced forms of attack before they are launched. Additionally, it provides the public and private sectors with opportunities for much-needed collaboration and information sharing to improve cyber attack defenses.
The strength of the CERT Analysis Center will come from contributions across the information technology community. SEI affiliate and visiting scientist programs provide an established model to integrate the contribution of diverse participants. These programs bring together members of academic, industry, and government organizations to address problems and meet common
needs. The center provides the means for private sector firms to collaborate with technical staff from the CERT Analysis Center on leading-edge information assurance research.
Research includes intruder tool analysis; that is, in-depth analysis of new and emerging cyber-attack methods in order to develop defenses and countermeasures that can be deployed before these new attack methods are widely used. Equally important is in-depth analysis of information technology vulnerabilities and malicious code in order to develop techniques that are effective at eliminating entire classes of vulnerabilities and entire families of malicious code.
Support the growth and use of global detection mechanisms.
Among the ways to gain a global view of threats are to use the experience and expertise of incident response teams to identify new threats and vulnerabilities. The incident response team at the CERT/CC and other response teams have demonstrated their effectiveness at discovering and dealing with vulnerabilities and incidents. Ongoing operation and expansion of open, wide area networks will benefit from stronger response teams and response infrastructures.
Similarly, it is important to encourage Internet service providers to develop security incident response teams and other security improvement services for their customers. Many network service providers are well positioned to offer security services to their clients. These services should include helping clients install and operate secure network connections as well as mechanisms to rapidly disseminate vulnerability information and corrections.
Support education and training to raise the level of security.
As noted earlier, the security of each system on the Internet depends on the security of all other systems on the network. The interconnectedness and interdependency of systems pose a serious threat to commerce.
The combination of easy access and user-friendly interfaces have drawn users of all ages and from all walks of life. As a result, many users of the Internet who have no more understanding of the technology than they do of the engineering behind other infrastructures. Similarly, many system administrators lack adequate knowledge about the network and about security, even while the Internet is becoming increasingly complex and dynamic. To encourage "safe computing," there are steps we believe the government could take:
In particular, support programs that provide early training in security practices and appropriate use. This training should be integrated into general education about computing. Children should learn early about acceptable and unacceptable behavior when they begin using computers just as they are taught about acceptable and unacceptable behavior when they begin using libraries. Although this recommendation is aimed at elementary and secondary school teachers, they themselves need to be educated by security experts and professional organizations. Parents need be educated as well and should reinforce lessons in security and behavior on computer networks.
Furthermore, the increasing need for such roles in organizations of many sizes and descriptions has led to assigning information security responsibilities to inexperienced personnel with little or no training. In the short term, the greatest need is for short "how to" and "what to be aware of" courses. In the long term, there should be undergraduate-level or master’s-level specialties in network and information security.
Support research and development in the areas of security and survivability of unbounded systems’ architectures with distributed control.
It is critical to maintain a long-term view and invest in research toward systems and operational techniques that yield networks capable of surviving attacks while protecting sensitive data. In doing so, it is essential to seek fundamental technological solutions and to seek proactive, preventive approaches, not just reactive, curative approaches. The research agenda should seek new approaches to system security. These approaches should include design and implementation strategies, recovery tactics, strategies to resist attacks, survivability trade-off analysis, and the development of security architectures. Among the activities should be these:
Conclusion
The Internet has proven to be an engine that is driving a revolution in the way individuals and organizations conduct business. Internet-based electronic commerce is growing rapidly and promises economic growth that would be difficult using any other means. Capitalizing on e-commerce opportunities, however, brings a new set of risks—risks that must be effectively managed to allow customers to have confidence in dealing with merchants, and merchants to protect themselves against losses from network attacks. Cybersecurity efforts focused on reporting and monitoring threats and vulnerabilities, education and training, and research and development are needed for US business to thrive on the Internet.
CERT and CERT Coordination Center are registered in the U.S. Patent and Trademark Office.