Skyport showed their SkySecure Center at Network Field Day 11 and I was impressed with what they showed. My problem with security is that it tends to be like sticking a finger in a leaky dike, or it is like poking at a bowl of JELL-O. Nothing you can do is really secure and you have to be 100% accurate or the bad guys get around your efforts. SkySecure is different in that it starts with fundamentals: verifiably secure hardware, BIOS, and the container for running a VM. Nothing special needs to be done to the VM, because security is provided by the platform.
The SkySecure hardware platform verifies that their platform is running known-good hardware, known-good firmware, known-good software, and known-good applications. This includes BIOS and I/O support firmware. Of course, you need to know that the original hardware support firmware is unchanged as installed. The VMs that run on their platform must communicate through their encapsulating systems, preventing one VM from bypassing external firewalls when communicating with another VM. Because the trust boundary for inter-VM communications goes through the SkySecure, it is easy to see what goes into and out of each VM. There is no question about whether some communications was able to bypass a firewall – everything is visible. This is different than any other security system I’ve seen.
SkySecure is able to log all communications with each VM, even to the extent of sending a PCAP file to external tools like Wireshark or Application Performance Management systems. This export capability facilitates troubleshooting of slow applications as well as verifying the content of any communications with a VM.
The Deployment Model
Their systems are more expensive than running your own hardware, so what makes this a compelling proposition? Well, what about applications that you can’t afford to have compromised, like DNS, a badge reading system, a video recording system, a corporate email server, or a credit card storage system? Another excellent use is to build out a DMZ, which is exposed to all sorts of external attacks. How about securing critical process control systems? The only thing their system needs is power and an Ethernet connection, making it easy to install the hardware.
The encapsulation of the VM basically builds a software-defined DMZ around the VM. System configuration and policies are easily constructed, making it easy to turn up a system. With their knowledge of many systems, SkySecure has policy templates that can simplify the deployment of an application that they have seen before.
I liked their ability to see all communications into and out of a VM as a way to determine what ports and protocols an application uses. For example, with healthcare clients, we often are faced with leaving security open because the vendors can’t tell us what ports and protocols their applications use. With SkySecure, it is easy to see all the communications and quickly build policies that match the application’s requirements. I could easily see building the core of a healthcare Electronic Health Record system around SkySecure.
The Bottom Line
I was favorably impressed with the SkySecure functionality. It is a new approach to server and application security that makes sense. They have done a good job of making this level of security easy to implement. I think they have a winner.
Wonder if they could do something like that for cloud server instances? Packet capture or preferably NetFlow would be pretty darn useful for managing the cloud. Feeding packets back could allow use of in-house tools to get some visibility into TCP delays, app mapping … I’m hearing that question more and more lately: “we have enough trouble keeping on top of applications in house, how are we going to manage them in the cloud?”