In Cloud We Trust?
Exclusive Q&A with Mike Gault, CEO of GuardTime
Apr. 8, 2011 12:00 PM
‘In God we trust,’ yet the currency of the Cloud is at odds with trust. Is it possible to trust applications that reside in a Cloud that seems so porous? Cloud Computing Journal's Jeremy Geelan sat down with GuardTime CEO Mike Gault, whose keyless signature technology is used to secure cloud hosting provider Joyent, whose customers include LinkedIn, Gilt, and Twitter.
Cloud Computing Journal: What is GuardTime all about? As one of the few folks in the Cloud space with an initmate knowledge of Estonia, having lived there for more than three years, I know that its origins lie there...
Gault: GuardTime was started in 2006, as you say, in Estonia. For those not familiar with Estonia, Jeremy, let's explain how unique it is in several aspects. First, it’s one of the most wired countries - the whole country’s IT infrastructure was built post-Internet. And so, everything is basically built using the Internet, including the banking systems. For example, 97% of transactions are done online and with this comes a very high knowledge of IT security by the general population.
The academics that invented the keyless signature technology were involved in the design of the infrastructure. They worked together with the local banks in Estonia. During this program, they realized that there was a fundamental problem that remained unsolved in IT security, which is trust. Why should I trust you with my data? Why should you have complete trust in your system administrator? The technology was built as an answer to these questions; as an alternative to the existing PKI technologies that are incapable of solving these specific problems.
Cloud Computing Journal: What happened to PKI (public key infrastructure)?
Gault: The U.S. has a long history in the 1990s and early 2000s with PKI as an IT security tool. It is very powerful for encryption and data privacy. It’s also very powerful for authentication but as a mechanism to prove the integrity and authenticity of data, it’s a horrible technology. It requires key management and trust authorities, both of which could be compromised and even a hint of a compromise would invalidate all historical records. In the end, it costs more to sign the data than it does to actually store the data. This is one of the reasons why PKI achieved very limited uptake in the U.S. for signing data.
Cloud Computing Journal: And then along comes Cloud computing around 2007.
Gault: Exactly. People realized that something fundamentally changed. It basically went from this data is inside my organization and I’m willing to trust my IT department to ‘this data is outside my organization and I can no longer trust the IT department or outsiders, the people that are now hosting my data.
Cloud Computing Journal: And where does GuardTime fit in?
Gault: GuardTime provides the infrastructure for delivering data signatures – the verification of which does not rely on cryptographic keys. That means that any data in the Cloud or outside the organization can now be signed, and the verification of the signature is based solely on mathematics. This means that you can give your data to your Cloud provider and to your clients, to your partners, or the data can be hosted inside the organization. And you have mathematical proof that the data has not been tampered with.
Cloud Computing Journal: So what’s the overhead on this? Do you see this being for specific types of data in communication?
Gault: The ultimate goal is for keyless signatures to be ubiquitous with data integrity. We have scenarios running up and down the stack from file system integration to system integration to application integration, all the way down to very specific application uses, such as non-repudiated e-mail.
We have, for example, a relationship with Joyent where we integrate into the Joyent Smart Data Center technology providing keyless signing services to their users. Application providers hosting applications on the Cloud stack will usually sign specific data items to ensure both data integrity and operating integrity. For e-mail systems, we have clients using the signatures to verify and validate both the sending party and the authenticity of the e-mail and all of its attachments.
Cloud Computing Journal: As you move data to the Cloud there are many different challenges. Applications have to be designed differently. Security gets pushed further and further away from perimeter-based approaches. How do you see security threats changing in the move to the Cloud?
Gault: Cloud computing is the best thing that has happened to information security in 50 years, because it’s forcing people to address the perimeter issue. Up until the existence of Cloud computing it was acceptable to trust the IT department internally. Now that the IT department is outsourced, all of a sudden people are asking the right questions about IT security. The focus must be on the data itself and not on the perimeter around the data. Cloud computing is forcing IT to ask similar security questions about activities inside the enterprise as well. Cyber defense and Cloud computing are actually closely related, because it becomes about who can you trust and what do you do if you can no longer trust the people inside your organization or institution.
Cloud Computing Journal: What are some questions people should ask a Cloud vendor as people weigh their choices between Joyent, Rackspace, or Amazon?
Gault: Most Cloud vendors will provide security by SLA (service level agreements). What these Cloud providers are saying is, “Please trust us. If anything goes wrong, then we’ll give you your money back.” That doesn’t really inspire much confidence. A much better way of providing security is via proof as in “if anything goes wrong we can provide your verifiable proof of what happened”. For cloud services the SLA is key and people should look closely at ones which require trust and ones which provide proof. Currently only Joyent can provide proof via SLA.
Cloud Computing Journal: How would you know that you got that mathematical proof?
Gault: Data is signed as it is stored in the cloud. That data could be application executables, logs or general data. The result of the signature process is a data signature which is stored either embedded inside the data item, with the data item or separately from the data item. To verify a signature you run a series of hash functions using the signature data and either check the result electronically or for the truly paranoid against a newspaper copy.
Cloud Computing Journal: Not every security threat can be solved through mathematical proof of data signing; there’s still application vulnerabilities and insider threats. What are some security questions you would ask?
Gault: Certainly confidentiality is a major question. Having the right tools in place to ensure that confidentiality is also being maintained is critical. So, some questions would be: What mechanism do you have to protect and securely deliver logs? What are you actually able to log? What activity are you recording within your Cloud? Can the integrity of those logs be proven regardless of when and where they are sent?
Cloud Computing Journal: Security logs are notorious for sitting around and not being accessed unless there’s a known problem. Now that we’re moving to the Cloud, are there better tools to make logs actionable in real-time rather than something that’s basically forensic?
Gault: The SIEM providers are looking at technologies to pull together information and leverage the Cloud resources as well. Users of the Cloud have to employ their own security measures based on their risk level and their regulatory requirements. What GuardTime provides is the ability to prove that those technologies have indeed worked as designed and planned.
It’s important that the technologies have been implemented properly when the logs are being created and that they’re being used properly to analyze and correlate the data into different sets of information. All those actions, including the policy definitions and admin changes associated with them, need to be signed. Because, one of the first things that an attacker will do after gaining access to a system, especially if they are an inside attacker, is to go into the logs and remove the entries that show how and where the access was gained and what has been done to modify the environment to leave the back door open. GuardTime can’t prevent those things from happening, but it can prove that those logs are, in fact, intact and haven’t been changed outside of the rules defined.
Reader Feedback: Page 1 of 1
Latest Cloud Developer Stories
Subscribe to the World's Most Powerful Newsletters
Subscribe to Our Rss Feeds & Get Your SYS-CON News Live!
SYS-CON Featured Whitepapers
Most Read This Week