Interviews

Virtualization security or ‘virtsec’ is one of the hottest topics in virtualization town. But do we need another abbreviation on our streets? Does virtualization require its own security approach and how would it be different from the physical world?

Different opinions fly around in the blogosphere and among vendors. Some security experts claim there is nothing new under the sun and the VirtSec people are just trying to sell products based on the Virtualization Hype. Some see a genuine need to secure new elements in the infrastructure, others claim that Virtualization allows new capabilities to raise security from the ground up and cynics claim it is just a way for the Virtualization industry to get a larger piece from the security budget.

So our editors Tarry and Kris set out to clarify the different opinions, together with the support of StackSafe, they organized a conference call with some of the most prominent bloggers, industry analyst and vendors in this emerging field.

On the call were Joe Pendry (Director of Marketing at StackSafe), Kris Buytaert (Principle at Consultant Inuits), Tarry Singh (Industry/Market Analyst Founder & CEO of Avastu), Andreas Antonopoulos (SVP & Founding Partner at Nemertes Research),Allwyn Sequeira (SVP & CTO at Blue Lane), Michael Berman (CTO at Catbird), Chris Hoff (Chief Security Architect – Systems & Technology Division and Blogger at Unisys) and Hezi Moore (President, Founder & CTO at Reflex Security)

During our initial chats with different security experts their question was simple: “what does virtsec mean?”. Depending on our proposed definition, opinions varied.

So obviously the first topic for discussion was the definition of VirtSec:

Allwyn Sequeira from Blue Lane kicked off the discussion by telling us that he defined Virt Sec as “Anything that is not host security or that’s not network-based security. If there’s a gap there, I believe that gap – in the context of virtualization – would fall under the realm of virtualization security. ” He continued to question who is in charge of Inter-VM communication security, or how features such as Virtual Machine Migration and Snapshottiting add a different complexity to todays infrastructure.

Andreas Antonopoulos of Nemertes Research takes a different approach and has two ways of looking at VirtSec “How do you secure a virtualized environment” and in his opinion a more interesting question is “How do you virtualize all of the security infrastructure in an organization” Andreas also wonders how to call the new evolutions “What do you call something that inspects memory inside of VM and inspects traffic and correlates the results? We don’t really have a definition for that today, because it was impossible, so we never considered it.” He expects virtualization to change the security landscape “Just like virtualization has blurred the line between physical server, virtual server, network and various other aspects of IT, I see blurring the lines within security very much and transforming the entire industry.”

Hezi Moore from Reflex Security wants to search for actual problems. He wants to know what changed since we started virtualizing our infrastructures. “A lot of the challenges that we faced before we virtualized; are still being faced after we virtualized. But a lot of them got really intensified, got much more in higher rate and much more serious.”

Michael Berman from Catbird thinks the biggest role of VirtSec still is Education, “..and the interesting thing I find is the one thing we all know that never changes is human nature.” He is afraid of virtualization changing the way systems are being deployed with no eye on security. Virtualization made it a lot easier to bypass the security officers and the auditors. The speed at which one can deploy a virtual instance and a bigger number of them has changed drastically regarding to a physical only environment, and security policies and procedures have still to catch up. “We can have an argument whether the vendors are responsible for security, whether the hypervisors about who attack servers. The big deal here is the human factor. “

Chris Hoff summarizes the different interpretations of VirtSec in three bullets:

• One, there is security in virtualization, which is really talking about the underlying platforms, the hypervisors. The answer there is a basic level of trust in your vendors. The same we do with operating systems, and we all know how well that works out.

• Number two is virtualized security, which is really ‘operationalization’, which is really how we actually go ahead and take policies and deploy them.

• The third one is really gaining security through virtualization, which is another point.

Over the past decade different Virtualization threats have surfaced, some with more truth than others. About a decade ago when Sun introduced their E10K system, they were boasting they really had 100% isolation between guest and host OS. But malicious minds figured out how to abuse the management framework to go from one partition to another. Joana Rutkowska’s “Blue Pill” Vulnerability Theory turned out to more of a myth than actual danger. But what is the VirtSec industry really worried about?

It seems the market is not worried about these kind of exploits yet. They are more worried about the total lack of security awareness. Andreas Antonopoulos summarizes this quite well “I don’t see much point in really thinking too much about five steps ahead, worrying about VM Escape, worrying about hypervisor security, etc. when we’re running Windows on top of these systems and they’re sitting there naked”.

Allwyn from Blue Lane however thinks this is an issue…certainly with Cloud Computing becoming more popular, we suggest to seriously think about how to tackle deployment of Virtual Machines in environments we don’t fully control. The Virtual Service Providers will have to provide us with a secure way to manage our platforms, and enough guarantee that upon deployment of multiple services these can communicate in a secured and isolated fashion.

Other people think we first have to focus on the Human Factor, we still aren’t paying enough attention to security in the physical infrastructure, so we better focus on the easy to implement solutions that are available today, rather than to worry about, exploits that might or might not occur one day.

Michael Berman from Catbird thinks that Virtualization vendors are responsible to protect the security of their guest. A memory Breakout seems inevitable, but we need to focus on the basic problems before tackling the more esoteric issues…He is worried about scenarios where old NT setups, or other insecure platforms are being migrated from one part of the network to another, and what damages can occur from such events.

Part of the discussion was about standardization, and if standardization could help in the security arena. Chris Hoff reasons that today we see mostly server virtualization, but there is much more to come, client virtualization, network virtualization, etc. As he says: “I don’t think there will be one one ring zero to rule them all.”. There are more and more vendors joining the market, VMWare, Oracle, Citrix, Cisco, Qumranet and different others have different Virtualization platforms and some vendors have based their products on top of them.

In the security industry standardization has typically been looked at as a bad thing, the more identical platforms you have the easier it will be for an attacker, if he breaks one, he has similar access to the others. Building a multi-vendor or multi-technology security infrastructure is common practice.

Another important change is the shift of responsibilities, traditionally you had the Systems people and the network people, and with some luck an isolated security role. Today the Systems people are deploying virtual machines at a much higher rate , and because of Virtualization they take charge of part of the network, hence giving the Network people less control. And the security folks less visibility

Allwyn Sequeira from Blue Lane thinks the future will bring us streams of Virtualization Security, the organizations with legacy will go for good VLAN segmentation and some tricks left and right because the way they use Virtualization blocks them for doing otherwise. He thinks the real innovation will come from people who can start with an empty drawing board.

Andreas Antonopoulos from Nemertes Research summarized that we all agree that the Virtualization companies have a responsibility to secure their hypervisor. There is a lot of work to be done in taking responsibility so that we can implement at least basic security. The next step is to get security on to the management dashboard , because if the platform is secure, but the management layer is a wide open goal, we haven’t gained anything.

Most security experts we talked to still prefer to virtualize their current security infrastructure vover the products that focus on securing virtualization. There is a thin line between needing a product that secures a virtual platform and changing your architecture and best practices to a regular security product fits in a Virtualized environment.

But all parties seem to agree that lots of the need for VirtSec comes from changing scale, and no matter what tools you throw at it, it’s still a people problem

The whole VirtSec discussion has just started, it’s obvious that there will be a lot of work to be done and new evolutions will pop up left and right. I`m looking forward to that future So as Chriss Hoff said “Security is like bell bottoms, every 10-15 years or so it comes back in style”, this time with a Virtualization sauce.

Listen to the full audio of the conference call!