From a security standpoint, basic server hypervisors have a lot of intrinsic strength. They work at a very low level within a given piece of hardware. They're hardened and task-specific, and the code base is relatively small. And it's a good thing, because the hypervisor enjoys a privileged degree of access to guest operating systems, especially via OS-native virtual machine tools, which allow the hypervisor all sorts of power. Compromising the hypervisor gives complete and total access to all of the data structures that comprise the system itself. But when we asked about hypervisor security, only 64% of respondents to our survey cited concern about this issue. That leaves a staggering 36%--greater than one-third of respondents--who have their heads in the sand. If a system runs code, it can be compromised, and if that code is running everywhere, there's a huge incentive to break it. There have been no fewer than 10 major hypervisor vulnerabilities disclosed this year alone, affecting a variety of platforms. Exploits range from remote code execution vulnerabilities (the most severe) to denial of service, and while VMware has yet to disclose a remote code execution vulnerability, it's only a matter of time. Earlier this year, for example, outdated source code for VMware's ESX hypervisor was posted.
We still see companies with a long way to go to integrate hypervisor awareness into their overall security mandates. The good news is that vendors have been preparing for this eventuality for some time, as we discuss in our full report. Also, about half of survey respondents (48%) have a hypervisor-aware security product in place. An additional 32% plan to adopt one.
A Shaky Virtual Stack