NC State, IBM Researchers Create 'Stealth' Hypervisor Security Tool

Researchers at NC State University and IBM have built a prototype security tool that operates in stealth mode to determine the security of a hypervisor so as not to tip off attackers.

The so-called HyperSentry software runs outside the hypervisor to verify in real time whether it has been compromised by malware or an attacker. Because the hypervisor handles multiple virtual machines within a physical machine, an attack against the hypervisor could compromise multiple applications and, in the case of cloud computing services, potentially multiple customers' systems and applications. The attacker then could steal user information, spread malware, or deploy the cloud's computing resources for other attacks.

Dr. Peng Ning, one of the researchers who developed HyperSentry and a professor of computer science at NC State, says the prototype tool measures a hypervisor's integrity without the hypervisor knowing it's being measured in case an attacker has wrested control of it. The software also has a full view of all elements of the hypervisor, including the internal memory of the CPUs running it, according to Ning. "It gives you peace of mind about the [cloud computing] system's integrity: At least you can be more confident that the system is better protected," Ning says.

Hypervisor attacks are rare today. "But if there is one, the consequences would be quite serious. Think about Amazon [cloud computing service] with so many machines running so many things [being attacked]," Peng says.

Sophisticated malware can be camouflaged to hide from security software that only views the memory where the hypervisor is stationed, and can modify pieces of the CPU and move the infected software where it can't be detected. HyperSentry is able to see where the hypervisor is located, even if it has been relocated by an attacker, according to the researchers.

"It makes sure it's really measuring the hypervisor," Ning says. Existing security tools can be defeated by hypervisor hacks, he says.

"The hypervisor is usually the highest-privileged software on the system. If the hypervisor is compromised, the attacker ... can change how the program is supposed to run. It can replace itself with a previously known version, and if a vulnerability exists, an attacker can come back and attack it again," he says.

Ning and his fellow colleagues built their prototype for the Xen hypervisor and will present their work next month at the 17th ACM Conference on Computer and Communications Security in Chicago.

The HyperSentry prototype uses the Intelligent Platform Management Interface (IPMI) on the server to set up the "out-of-band" channel that separates the tool from the hypervisor. It then invokes a System Management Interrupt (SMI) to trigger HyperSentry. "We're also working on a version to measure the Linux KVM" virtual environment, Ning says.

HyperSentry ultimately will become an open-source offering. The researchers are looking at other applications of the technology, as well. "We are having conversations with IBM and Red Hat to see if it can be applied to enterprise systems management. So if this works out, hopefully in two to three years you will see HyperSentry used in an enterprise environment," Ning says. "HyperSentry is currently shaped for server platforms."

NC State researchers earlier this year built another hypervisor security tool called HyperSafe that blocks any new code -- namely malware -- from getting into the hypervisor and restricts alterations to the hypervisor's code. Ning says HyperSafe is complementary to HyperSentry.

Have a comment on this story? Please click "Discuss" below. If you'd like to contact Dark Reading's editors directly, send us a message.