1 July 2005
Best practices for control system security.
By Ernest Rakaczky
Intrusion detection systems (IDS) represent a critical piece of security infrastructure you should implement whenever you connect critical systems, such as industrial process control systems, to TCP/IP-based local and wide area networks. They can detect network activity such as hacking attempts, virus and worm attacks, and other potentially threatening traffic capable of wreaking havoc on your control system. The technology behind them is simple—detect a threat and alert you. Today, new-generation IDS, also known as intrusion prevention systems (IPS), are not only able to detect threats, but they can mitigate them by blocking the traffic from entering your network.
Intrusion prevention is a new technology category focusing on taking a proactive approach to IT and control network security by preventing attacks on multiple network resources, as opposed to similar technology that merely detects and reports on attacks that have already taken place. Intrusion prevention is gaining visibility in corporate and government organizations due to the inherent limitations in existing security technologies. The significant financial loss organizations experienced is valid proof. Think of it as the logical follow-up to signature-based technologies, such as intrusion detection and anti-virus, and to network-oriented protection solutions, such as firewalls.
Traditional security products focused on the biggest threats emerging as computer networking. Corporations adopted e-mail and Web applications and purchased products to solve the security issues inherent in these technologies, namely perimeter protection (firewalls), network protection (network-based intrusion detection), and file-based security (anti-virus). But these technologies don't address new attacks that ride over existing protocols to attack applications or new content-based attacks that attack systems before vendors are able to release and distribute signatures and other countermeasures.
How can your organization make informed decisions when choosing intrusion prevention products? Here are some technology best practices for intrusion prevention solutions that might help.
Any organization that intends to protect itself by using intrusion prevention technology should consider a number of factors when evaluating products that address the organization's defined security requirements. Take care to choose solutions meeting corporate security, manageability, and flexibility requirements, lest the solution be a partial one, or worse, introduce a significant management burden that overshadows the security benefits.
1. Host-based protection
As more companies adopt technologies like high-speed networks, switching, and end-to-end encryption, providing desired security at the network level becomes a major challenge.
The best place to enforce security is at the desktops and servers, where you perform the actual work and where the potential for damage is greatest. Consider the value of only a host IPS solution that is not signature or role based dependent. In the control environment, one of the key current challenges is the overall deployment and execution of current security vulnerability patches for given operating software.
2. Real-time prevention decisions
To ensure the highest levels of security and minimize the ability to bypass the security policy on a host, you must intercept application calls at the kernel level, where you determine their adherence to policy. You can easily bypass implemented solutions by replacing shared libraries or analyzing system audit logs. An effective intrusion prevention strategy includes preventing violations in real-time, rather than noting attacks or system changes after the fact.
3. Defense in depth
To completely enforce a company's security policy, intrusion prevention must intercept all major points of communication between applications and the underlying system.
Network control must limit client/server communications at the port and protocol level, as well as hosts for permitted communications. File system controls must allow or deny read and write access to folders and files on an individual and group basis. Registry controls must prevent the overwriting of important registry keys that control how the system and other applications operate. And, COM controls should restrict inter-process communication to allowable access.
Attacks have multiple phases: exploiting network and application-level weaknesses, replicating and distributing themselves, and making unauthorized changes to the system. A complete intrusion prevention strategy must protect systems from all these phases, so if a new class of attack releases, you can thwart it at one or more of the stages.
4. Real-time correlation
Correlation deployed at the agent provides a level of accuracy on prevention decisions that does not exist with signature matching approaches. Correlating sequences of events within the context of an application's behavior eliminates the potential for false positives.
Correlation at the enterprise level enables security to be adaptive. By correlating the events on distributed agents, you can dynamically update intrusion prevention policies to prevent propagation of malicious code, thus preventing widespread damage to numerous resources.
5. Behavioral approach
The intrusion prevention approach must enforce appropriate system and application behavior to ensure the security implemented is proactive, not reactive. Solutions that rely on signatures only provide security to the release of the most recent signature update.
Every corporation is unique in how it configures and manages details of its systems and corporate applications. Intrusion prevention solutions considered must permit the policy customization and creation to accommodate unique applications and implementations. The solution must support automated policy creation to ease the management burden of creating policies by hand.
7. Ease of deployment
The intrusion prevention strategy should minimize the personnel overhead associated with agent deployments. Solutions considered must provide out-of-the-box functionality to allow for rapid deployment of the desired security policies, and must allow for roll out of new and custom policies as needed without additional intervention at the host level. The solution must support Web-based deployment, and allow for easy integration with standard corporate software distribution mechanisms.
8. Centralized event management
All events the agents generate must roll up into a centralized repository from which alerts and reports may be generated. Solutions considered must support standard alerting interfaces, such as SNMP, paging, e-mail, and flat files. They should also allow for custom interfaces to the alerting system to easily integrate with corporate systems.
9. Platform coverage
Solutions must provide coverage for the key operating systems the corporation wants to protect. In light of recent attacks, like Nimda, which target multiple hosts, the same management and enforcement paradigm must apply to desktop and server-based systems.
To ease policy management, policies must be definable centrally and automatically distributed to agents on a configurable interval. They must also be exportable for replication and archive purposes.
Behind the Byline
Ernie Rakaczky is director of education at Invensys Systems Canada, Inc., in Dollard-des-Ormeaux, QC, Canada.
Top plant security myths
By Bryan Singer
As awareness of the issue of plant floor security continues to grow across the industry, one question I often hear is "what are the real risks to a manufacturing environment?" When most of us think of security, it usually invokes visions of viruses, hackers, and worms. While these threats are certainly real, it's important to recognize the vast majority of security breaches come from within a company's walls, through acts by employees and flaws in security procedures. Along with the growth in use of open systems technologies on the factory floor, including Web browsers and standards-based networks, comes a whole new host of vulnerabilities that didn't exist before, at least in the plant-floor environment.
With leaner operations and more integrated plant information systems, keeping information and control systems secure and reliable requires manufactures to stay keenly aware of everything happening on the plant floor. Security breaches occur daily in plants, and many of them are simply the result of faulty procedures or poor personnel oversight. While employing the right technology is important, it is also key to effectively manage people in this environment to ensure optimum plant-floor security. Below are some of the most common security misconceptions throughout the industry.
Common security misconceptions
Myth: Technology solutions can answer all of the problems.
Reality: Most security mishaps are the result of a plant's own people, policies, and procedures. The best technology in the world isn't going to make a facility secure if there aren't properly trained people to use it correctly. To execute a successful strategy, it's imperative manufacturers employ trained personnel and enforce policies that assign responsibility to individuals who can then be held accountable for any security incidents.
Myth: If your organization has an IT department, your plant is secure.
Reality: This is sometimes true, but not often. There are significant differences between security in an IT environment and security in an industrial automation and controls setting. In a plant environment, avoidance of downtime is crucial, and if there is a failure, immediate action needs to take place to restore production and minimize losses. IT departments don't have the same sense of urgency for repair of the manufacturing systems and therefore, by relying solely on your IT department, your plant could be at even greater risk.
Myth: Security is a significant company expense.
Reality: The misuse of technology results in significant expense, not security itself. Having properly trained personnel in place who know how to use technology correctly helps ensure you're using your investment without overspending. Contrary to popular opinion, it is possible to employ skilled people who are capable of following defined practices and can carefully, accurately, and efficiently apply technology.
Myth: The software vendor is responsible for patch certification.
Reality: While it is the responsibility of vendors to test their products against general patches and give guidance on patch management, it is the responsibility of the user to create internal labs and test the compatibility of patches against their own environment. To do this correctly, you must develop the appropriate security architecture for your controls environment. In simple terms, the software vendor can't control the environment and therefore can't account for all of the variables.
Myth: Preventive and detective measures are enough to keep a plant secure.
Reality: Regular patch testing and virus updates alone aren't going to effectively manage security. These measures only work if you do them before a virus hits, and preventive and detective measures can't substitute for trained people and sound security policies. The best solution is to have better defenses and more preventive measures in place.
The technical reports from ISA and other upcoming documents can be excellent resources to help ensure a secure plant. After reviewing these reports, it's important to conduct a risk analysis of the plant's control system to identify potential security risks and assess any potential problems. Assemble an internal team involving the major business units and develop a comprehensive security plan. If your company is not informed on security risks, engage experts who can thoroughly educate and inform management on this topic. Once you complete a thorough examination of your facility and procedures, take corrective actions to help ensure maximum plant reliability and security.
Behind the Byline
Bryan Singer is a senior business consultant at Milwaukee-based Rockwell Automation, and chairman of the ISA SP99 Committee.
Return to Previous Page