Monday, October 13, 2008

A horse's ass approach to virtualization security

The interest and excitement around virtualization is palpable. However, it seems like the security approaches in this area are similar to the constrains that a horse's ass put on the space shuttle design.

Virtualization security solutions today primarily focus on protecting the virtual OS, the virtual networks, or the hypervisor software itself. More specifically, most current virtualization security technologies are focused on preventing hypervisor root kits, providing intrusion detection, anti-malware, anti-virus, network security, etc. In the physical world, this is similar to individually protecting hardware, operating systems, and the networks that connect them. That is, the focus is mainly on protecting infrastructure and perimeter, not data. Protecting that data, however, should be the single most important aspect of virtualization security.

Here is why: Any execution environment requires four elements: devices/hardware/OS, networks, applications, and data. With the advent of virtualization, physical devices/OS are being replaced by flexible, on-demand virtual “devices,” networks are being virtualized and applications are being streamed down from virtual environments. Therefore, the only remaining “constant” element is the data itself - which also has a longer lifetime than the ephemeral virtual environment. While protecting the virtual infrastructure is important, I believe the primary focus for protection should be the data – the true IT asset.

Virtualization is a game-changer for computing and has forced the IT world to rethink its infrastructure; now virtualization security has to be rethought as well. An information-centric approach to persistently protecting the data itself is the only way to really benefit from virtualization and keep the data truly secure.

Or thinking about it another way - why was Google's approach to navigate the web using search better than the initial Yahoo approach of hierarchical mapping? Coz Yahoo was mapping an old yellow-book approach to managing data, while Google took advantage of the new medium.

I shall try and elaborate on my thoughts in upcoming posts...

No comments: