With high profile VMWare vulnerabilities just hitting the news its easy to find some mainstream articles covering the subject. This post isn't about hypervisor rootkits (because were all tired of hearing about that), but more about the assumption in corporations and academia that (virtualization == security). This is just plain WRONG. Virtualization environments are extremely complex pieces of software - and with complexity comes insecurity. In fact I would venture as far as to say that by default (virtualization == insecurity); running two operating systems within the same machine just creates more attack surface. Considering the high degree of interaction the host and guest OS must have you inherently create greater possibility of vulnerability then if they were on separate hardware. And just because VM's are easy to create and re-create doesn't mean they shouldn't be secured as well. As we have seen from this latest VMWare vulnerability, theres always the possibility your guest VM can compromise your host OS. It should also be noted that once the host OS has been hijacked ALL of your guest VM's should be considered compromised and untrusted. In order for the attacker to completely own your virtualization environment he/she has to know exactly what host OS is being used. There needs to be more fool-proof research into this area before wide spread panic can begin. There will also hopefully be more utilization of the host OS/virtualizer as an Virtual IDS (VIDS) of sorts - to tell us when our virtual machines have been compromised. This use hasnlt been explored enough in my opinion.
Now its true some virtualization technologies were designed with security in mind and others were meant to increase efficiency and productivity of hardware. This fact should be noted when trying to decide which virtualization strategy to use. But companies should also be aware of the security issues they may be introducing by improperly implementing a virtualization strategy as they may be causing more harm then its worth.