Alan: What do you think about future approaches to security such as a dumb terminal approach (i.e. Citrix or VNC in a world with infinitely fast bandwidth and infinitely small latency)?
Dino: I think we are moving towards a Web-based thin-terminal world, whether we like it or not. Once consumers realize that when their data is stored in the cloud, that they never have to worry about losing it, then they will begin to prefer it. If providers give users enough options such that they believe that they are as in control of their data as they are on their own system, they will have faith in it. This means allowing users to encrypt their data so that even the provider cannot see the file names or their contents.
Alan: That’s the thing though, is it a better solution? I’ll be the first to admit that I use Gmail because it’s so convenient to have “email anywhere.” With that said, I’m sure there’s stuff I’ve emailed via Gmail that I probably wouldn’t want anyone else to read.
I assume Gmail has redundant storage, but what happens if their hard drives crash, or my Internet is dead? If I was a hacker, wouldn’t it always makes more sense to try to exploit Gmail (and get millions of credit card numbers) as opposed to my own personal computer and get one person’s financial info?
Once the world moves to SSD, I’d predict that an individual user would have similar levels of reliability. Plus any sort of encryption Google could do, an individual could do on his home system (if not better encryption given that he’d be able to dedicate CPU resources toward a single user). Do you encrypt everything on your personal desktops and notebooks?
Dino: I’m also a huge fan of SSDs. I love their silent, fast, and reliable operation. As for data confidentiality, I use full-disk encryption and power down as often as is conveniently possible. The existing attacks against FDE require access to a powered-on or very recently powered-off system.
Alan: What about secure hypervisors?
Dino: So far, secure hypervisors have been used to protect the hardware business model from the users and owners of those systems. Systems such as video game consoles use secure hypervisors to prevent the owner from tampering with it. I would love to see software manufacturers provide a secure hypervisor to protect my data.
Alan: As would I, provided it was developed by a talented company and was reasonably priced. In the 90's, security researchers had to deal with the threat of polymorphic viruses which could elude many signature-based anti virus tools. What do you think the challenges in the next 2 years will be?
Dino: Signature-based anti-virus is an optimization that we have confused for a solution. The challenge over the next few years will be developing and deploying systems that are able to detect and prevent unknown exploits and malware. The highly profitable business model of signature-based anti-virus subscriptions discourages those companies from developing better and more generic solutions to the problem. That, however, leaves room for start-ups to innovate in this space.