Q: Patrick, I understand that IBM's Threat Protection System is capable of both detecting and preventing even unknown attacks, including those utilizing advanced malware. How does that function -- and what are its capabilities compared to competing systems?
Patrick Vandenberg: We know in today's advanced threat landscape that it is an incomplete strategy to rely on point solutions or focus on a single point in the attack chain. The continual stream of advanced attacks on enterprises requires a coordinated strategy to disrupt the lifecycle of these attacks. IBM's Threat Protection System is designed to do this across three critical areas: prevention, detection, and response, as follows:
- Prevent even the most sophisticated attacks. Although many in the industry have redirected focus on detection, many have done so at the expense of prevention. But prevention remains crucial to effective enterprise security. Real-time prevention is essential to stop advanced attacks from penetrating the organization. This is no easy task. But with behavioral-based capabilities that can protect against unknown and zero-day attacks, the IBM Threat Protection System can block the initial phases of an attack at the endpoint and network. An innovative new product called Trusteer Apex disrupts exploits leading to advanced malware on users' computers, while IBM Security Network Protection (XGS) prevents attacks from reaching vulnerable hosts; they also work in tandem to block attackers from establishing external control channels. Complimenting endpoint and network prevention is real-time policy enforcement of data access provided by IBM Guardium Data Activity Monitoring.
- Detect stealthy threats across the entire infrastructure. Even the strongest immune system cannot prevent 100% of invaders from getting inside, making it essential to quickly detect active threats hiding across today's complex IT environments. We solve this problem with data. Working as the central nervous system of our approach, the IBM QRadar Security Intelligence Platform is able to combine massive amounts of data from network traffic, user behavior, security events, and numerous other sources to automatically identify unknown or previously undetected threats. Real-time analytics find stealthy attackers lurking within the enterprise, while pre-attack analytics predict and prioritize security weaknesses before someone else does. This is the meaning of Security Intelligence.
- Respond continuously to security incidents. Finally, in the event of a successful security breach, it's important to quickly minimize its impact, understand exactly how the intrusion occurred, and learn from findings to prevent another similar incident. This rapid investigative capability is provided by IBM Security QRadar Incident Forensics, an offering that can scale investigative activity to security teams to quickly retrace breaches step-by-step, often in hours instead of days when time is most critical. This new solution, coupled with the expertise of our IBM Emergency Response Services and real-time incident response with automatic quarantine of non-compliant endpoints from IBM Endpoint Manager, helps organizations mount a strong and adaptive response to future occurrences of attack.
As previously mentioned, siloed security activity is an insufficient defense against today's cyber attacks, so the IBM Threat Protection System has prioritized integrations across the prevent, detect, and respond pillars to not only share and correlate information in a security analytics platform for security insights, but can also act on these insights with integrations that can update security control policies for immediate protective response, such as the "right-click" integration from IBM Security QRadar to IBM Network Protection XGS. This integrated system of prevention, detection, and response provides a unique ecosystem across hundreds of vendor offerings for a greater disruption of cyber attack activity.
Q: Chris, in your most recent X-Force Threat Intelligence Quarterly, it talks about the Internet of Things and whether ubiquitous connectivity means less security. Can you give me some insight into that?
Chris Poulin: The IoT encompasses a broad range of devices, including smart home electronics and appliances, consumer wearables, connected vehicles, implantable medical devices (e.g., pacemakers and insulin pumps), manufacturing and energy and utility systems (e.g., assembly robots, pumps and gates), and a myriad of others. One problem is that not all devices operate the same and, while some may run on general operating systems -- such as Google Android, Apple iOS, Windows Embedded, Blackberry NGX, and various Linux-based distributions, such as Tizen and Ubuntu -- others run on real-time OSes. Some are controlled and monitored through Web portals and mobile apps, while others may not have a human interface at all. The problem is that the security community at large is treating the IoT as a monolithic entity with one silver bullet. The reality, though, is there is no reductionist, unified theory of security for the IoT. Rather, we need to divide devices into functional categories with a set of appropriate security controls. Those include:
- Trusted firmware and rapid updates. Many IoT devices are cobbled from various parts, with the hardware, firmware, and software assembled who knows where and touched by who knows whom along the way. And when a vulnerability is discovered, it needs to be patched as soon as possible. This poses a few challenges because IoT devices aren't guaranteed to be connected to the Internet at any given time, and many tether over mobile channels as they travel with the consumer. IoT devices may be in the middle of performing a critical task and firmware updates may interrupt -- or worse, corrupt -- the operation of the device. Over the air and over the wire updates over untrusted networks create the risk of an attacker tainting the firmware or app. An integrity checking mechanism is imperative, and possibly a mechanism to revert back to a known good firmware version without interrupting the device's operation (e.g., a vehicle in transit.)
- Encryption for data and commands. Both at rest and in motion, data may contain private consumer data or metadata. IoT devices must preserve the privacy of consumer data, such as name and geolocation. IoT devices may even provide transitive access to consumer mobile devices and infrastructure -- e.g., an attacker breaks into a connected vehicle to which the driver has paired their mobile phone, or an attacker breaks into a connected toothbrush and uses that point of presence to attack the rest of their home network. Attackers can also inject commands onto the local network to induce a device to perform in a way it wasn't intended. Encryption and message integrity help solve these problems.
- Device identification and authorization. Another method to stop bogus control commands, or even connection of rogue hardware, the ability to uniquely identify device beyond simple IPv6 addresses, is important. There must be a method for the device to register and identify itself without user intervention. For example, the electronic control units in a connected car must be authorized to coexist and send control commands on the controller area network bus.
- User authentication where appropriate. On the user side, such as a mobile phone or Web portal, the authorized operator of the IoT device must be positively identified and authenticated. This is a standard security requirement, but worth mentioning if for no other reason than to draw the distinction between user interactive devices and those that intercommunicate without user intervention.
- Policy enforcement. What is appropriate between IoT devices? Should the window control module be able to turn off the engine? The wheel speed may control the volume of the sound system, proof that inter-network communications is useful, but setting policies on what should be able to send which control messages to other devices is important, although complex in a broad IoT ecosystem. Policy enforcement is tied to device identification and, optionally, user authentication.
- Behavior monitoring and intervention. Because of the complexity involved in designing interoperation polices, and given that some policies may be dependent on external, analog factors, such as human behavior, it makes sense to have monitoring devices profile behavior and identify anomalies against the baseline. For example, general policies can be set in a connected car (window control module should not be able to send control messages to the engine). However, different drivers will exhibit varying driving patterns. Some may drive with one foot on the gas and the other on the brake while others may be hyper-aware of traffic movement ahead and use the brake sparingly, preferring instead to let up on the gas in advance of slowdown or gear down in a manual transmission.
On the maker side, many come from the hardware world, where software is foreign. When seasoned software developers write code for IoT devices, they often are not experts on handling conditions where entropy in the real world intervenes. The consequences of vulnerabilities in traditional software are generally loss of money and time, exemplified by the huge number of retailer compromises in 2014; whereas, the consequences with IoT devices are often consumer safety. IoT products are being rushed to market by makers who may not be experts in both consumer safety and software security.
Q: Chris, that same publication focuses on who is the top offender for malware hosting? For those who haven't read that article yet, how about giving us some bullet points on that. Who is the top offender?
Poulin: IBM X-Force researchers continuously track sites that contain malware and store the information in our IP reputation database. We analyze this data to establish a baseline of the sources of massively distributed malware: countries where malicious links are most often hosted, based on our research, as well as the geographic distribution of botnet command-and-control (C&C) servers. When it comes to the top countries hosting malware, the United States has historically topped the list by a large margin, followed by Germany and China, vying to place each year. For countries hosting botnet C&C servers, we find a similar pattern, with the United States in the lead, with the Russian Federation, Republic of Korea, China, Germany, and the United Kingdom with strong showings.
However, that's on a straight numbers basis. It's not surprising that the countries with the greater numbers of technology users and service providers figure higher in the rankings. Consequently, we decided to normalize the figures based on the ratio of IP addresses as a percentage of total IP-addressable systems in the corresponding country. The result is that the U.S. moves out of the top 20 countries for hosting malware -- down to number 25. Hong Kong, Lithuania, and Bulgaria now appear in the top three positions. When normalizing the data for C&C server contaminations, the U.S. moves out of the top 20 countries for C&C servers -- down to number 28. This time, the Russian Federation only moves from second to third. Lithuania comes in first by a large margin, and Belarus, Slovakia, Ukraine, Turkey, Thailand, Hong Kong, Hungary, the Czech Republic, and Poland all appear above the average, which is just slightly less than two contaminated systems per one million.
In short, Lithuania leads the pack in C&C server contamination and comes in second for malware contamination, while Hong Kong comes in first for malware contamination.