Protecting one's data in the cloud, post Snowden
Andy Thurai, chief architect and group CTO, application security, API & Big Data, Intel
The ripple effects of Edward Snowden's actions are going to remain for a long time to come. If you haven't done so already, I suggest you read about the NSA surveillance programs PRISM and XKeyscore before you continue this article.
Essentially, these government programs are creating nervous times for my Canadian, European and APAC customers who are using U.S. cloud providers. Given the very stringent restrictions in those regions on data residency and data privacy to protect their citizens' sensitive information, this “guilt by association” mentality is going to hurt most corporations that move their data cross-boundaries.
But one thing is for sure, these are the programs that are exposed because someone came out in the public. Just because a specific country's cloud provider hasn't been accused yet (or not found guilty) doesn't necessarily mean that they are not doing the same thing. There is a chance that they might be doing it and have not been caught yet.
Unfortunately, cloud enthusiasts spent years alleviating the fear of moving data to the cloud. Those days, the fear was about hackers and disgruntled employees/partners accidentally or willfully exposing their data. Now, they need to fight an uphill battle of convincing certain organizations not about hackers, but about governments. It is probably not going to be pretty.
On top of that, there are foreign governments freaking out about this, such as Germany. Though the intent is good, it still is an after-effect. You can sanction cloud providers all you want, but the lost brand reputation and the associated revenue loss will hurt you more than the sanction and the lost compensation you might get out of it.
There are reports suggesting that U.S. cloud computing industry could lose as much as $35 billion over the next three years. A recent Cloud Security Alliance survey gives an early indication to this. About half of the survey's non-U.S. respondents say they are going to move away from U.S. clouds. But where would they go, and how can they be sure where they are moving is better and not jumping from the frying pan into the fire?
For foreign entities, such as European corporations, choosing a cloud provider in the U.S. seems to rank high on their list. This is because there is a large number of cloud providers in America that provide mature and wholesome solutions, and their pricing is very competitive. Until now, all one had to worry about was the data residency (or privacy) issue, but now you need to worry about a U.S. government agency spying on your data.
Well, here is a novel idea – don't send the sensitive data to the cloud. Send either encrypted garble or tokens that look like the original.
When you encrypt the data before it leaves your perimeter, you control the fate of your data. You get to choose the encryption key, the specific algorithm, key management, etc. This means that you get to control who gets to see what, when and how much. This is really important. If anyone needs to see your sensitive information, they have to come to you rather than going to your data sitter. You control the keys to your kingdom.
The second option that is gaining momentum recently is tokenization. In this case, you take the original sensitive data out, store it in a secure vault, and replace it with a random token that looks, feels, and acts like original data. The premise of tokenization is that what is not there cannot be stolen. Let hackers and governments have fun with it without knowing they have fake data.
This is where tokenization/data privacy solutions can help. It intercepts any message that goes out, in any enterprise messaging format (any structure or unstructured data), and scans for the sensitive data that is in the message, removes this information (such as credit card, personal information, health records, financial records, etc.), stores them in a safe place, and replaces them with a random data that is formatted exactly as the original data. The only agency that can co-relate the token to the original is you and no one else.
Interestingly enough, these solutions are touchless. This means you don't have to touch or modify any of your existing applications. You drop these tokenization/data protection gateways in the line of traffic, regardless of the type of traffic or type of message/data, and it will automatically sense the data based on pre-defined policies and work its magic.
But, remember, you still may have to be ready when a government agency comes knocking on your door asking for this information.
Andy Thurai is Chief Architect & Group CTO for the Intel unit responsible for Big Data, Cloud/Application security, SOA, API, and Mobile middleware solutions. You can follow him on Twitter: @AndyThurai, or on his blog.