Data-centric Encryption Definitely Protects against Data Attacks
In the aftermath of last week’s data breach at healthcare provider Anthem there have been inaccurate and misleading claims around the ability of strong encryption to protect data against attacks such as this.
In order to understand these issues around strong encryption, we need to first understand Internet protocols. The Internet protocol suite that’s sometimes known as “TCP/IP” defines the communication protocols used in most of today’s computer networks. One notable feature of TCP/IP is how it abstracts the functionality of a network into four layers that we think of as comprising a “stack,” in which information only gets passed between adjacent layers of the stack. A similar conceptual model of an “encryption stack” can help understand both the capabilities and limitations of different approaches to encryption.
The TCP/IP stack as defined in RFC 1122 and RFC 1123 comprises four layers: Application, Transport, IP and Network Access, as shown in Figure 1. As the arrows in Figure 1 suggest, information is only passed between adjacent layers of the TCP/IP stack. So a process running at the Transport layer can pass information to a process running one layer away at the IP layer but not to one running two layers away at the Network Access layer.
Figure 1. Conceptual model of the TCP/IP protocol stack.
Similarly, it can be useful think of encryption as taking place either relative to or at different levels in the TCP/P stack, possibly creating a notional “encryption stack” that closely parallels the TCP/IP stack.
The analogy to an “encryption stack” can be made as follows: TLS encryption, for example, operates between the Application layer and the Transport layer. IPsec operates at the IP layer. Link encryptors encrypt at the Network Access layer. Full-disk encryption operates below the Network Access layer.
There are good reasons to encrypt at different places in the TCP/IP stack, but when you encrypt at a particular location, the encryption only protects against threats that target layers at or below where the encryption takes place. For example, if you protect data with full-disk encryption, it protects the data while it’s on the encrypted disks, but when the data leaves the disks when it gets passed up the stack to a process running at the Network Access layer, that particular form of encryption no longer protects it.
Or if you’re using TLS to encrypt data between the Transport and Application layers, the TLS encryption will protect against attacks that target the Transport layer, the IP layer and the Network Access layer, but it won’t protect against attacks that target processes running at the Application layer. Once data that’s encrypted using TLS gets passed up the stack to the Application layer, the TLS encryption is no longer protecting it.
And if you encrypt at the application layer, then the encryption will protect against attacks that target any layer of the TCP/IP stack at all.
So if Anthem was doing encryption at the Network Access layer, it’s certainly reasonable to assume that the hackers that penetrated their network would have easily been able to bypass this encryption. And in this case, it’s true that strong encryption would not have thwarted the hackers.
But if Anthem had been encrypting at the Application layer, then it’s reasonable to assume that the hackers would only have managed to extract cipher text. If this is the case, then the encryption would definitely been able to thwart them.
This is why data-centric encryption at the Application layer is a more comprehensive solution to data protection -this approach protects against attacks that target any level of the TCP/IP stack, with hackers getting nothing more than cipher text from any of the networks that they manage to penetrate.
Newer data-centric encryption approaches like format-preserving encryption (FPE), and methods like secure stateless tokenization (SST), can protect data in such a way that decryption back to a live and vulnerable form isn’t necessary for most applications. Data-centric security removes the value from data that hackers might get through data breaches because it protects data at rest, in transit and in use. So when an attack happens, the attackers get nothing of value. In fact, Aetna and other organizations can use the data in its secure form for all existing business processes.
This is a unique difference to the encryption most journalists are writing about. Format-preserving encryption provides a new model for data security which has been adopted by payment processors, financial organizations, retailers, enterprises, and data aggregators today. NIST understand the value format-preserving encryption offers to data breach mitigation as demonstrated by its inclusion in the NIST Special Publication 800-38G: Recommendation for Block Cipher Modes of Operation: Methods for Format-Preserving Encryption (draft).
What defenses did Anthem actually have in place? Right now, that doesn’t seem to be well known, but it will almost certainly be widely discussed over the next few weeks. And it certainly seems likely that a blanket statement that strong encryption would not have prevented hackers from exploiting Anthem will turn out to be false, or at least false in any meaningful way.
Chief Security Architect, Voltage Security