A company is only as secure as the weakest link in its security posture. Finding and fixing security gaps is clearly a constant struggle for security professionals.
Security has traditionally been implemented with a layered multi-vendor multi-solution approach (with different vendors for endpoint solutions, gateways, clouds, etc.). That increased visibility but didn’t fix the problem. There are simply too many layers, too many vendors and way too much complexity for security teams. Cybersecurity consultants offer specialized gap analysis services to find the holes in an organization’s security fabric that attackers are anxious to exploit.
But still, the gaps persist. Take encryption, for example. Ideally, encryption should be invisible – a service of the underlying infrastructure, similar to how it is on the Mac on which I’m writing this article right now. Turning on hardware-grade disk encryption changes virtually nothing about how I do my work. That’s how encryption should work.
Unfortunately, nearly every piece of IT infrastructure (cloud, storage, network, application and database) has its own encryption scheme – if they even have one at all. This creates incredible complexity for IT, developers and even users. It’s challenging to determine if the data perimeter established is contiguous and complete. Is the data currently being encrypted? Is it double-encrypted? What level of encryption is being applied? Where does decryption take place? Are access policies consistent? This mind-boggling complexity inevitably creates gaps and results in unnecessary spending on layers and layers of security protection.
Even worse, there’s a massive hole in the data perimeter present in virtually every host operating today. All of those myriad encryption solutions address only two of the three states of data: data at rest (storage) and data in motion. That’s fine on a Mac where I’m the only user, and my concern is only about stored data being revealed if I physically lose my machine. But in a cloud or data center, it’s data in use that may be continuously exposed. Data is useless if it can’t be used and manipulated during runtime. Yet, during this runtime state, data becomes most vulnerable because it must be unencrypted for processing and remain in the clear in memory. Insiders or attackers who access a host can implicitly help themselves to highly valuable data in memory.
Security is undermined when protection is applied inconsistently or when only some assets or attack surfaces are protected while others are left unguarded. The lack of any data-in-use protection undercuts the strength and value of an organization’s entire set of data security controls. Consider the need to use cryptographic keys to decrypt data. Keys are subject to the same exposure as the data they’re designed to protect. To be used, they must sit in memory. A lock has far less value when the keys are generally available – or hanging on a peg right next to the lock.
Virtually every major cloud vendor has deployed new hardware to close this gap and limit the shared risk and responsibility they carry with each of their customers. This hardware provides the ability to completely encrypt data in use, cutting off implicit or explicit data access to insiders or attackers and effectively extending the data perimeter around data in memory and data being processed. With these confidential computing technologies, encryption keys can be completely locked down. Previously, this kind of hardware isolation wasn’t available except through complex, expensive hardware security modules. Now, it’s virtually ubiquitous, offered by the major cloud vendors at no additional cost.
“Security is undermined when protection is applied inconsistently or when only some assets or attack surfaces are protected while others are left unguarded”
While widely available, these security features are not commonly used because they usually require applications to be rearchitected and processes restructured. This hurdle puts this technology out of reach of most enterprises. The incredible promise of both closing a significant gap and consolidating complexity into a single, ideally invisible data security platform (as in my Mac) remains unfulfilled.
New software addresses this challenge. Invisibly integrating robust, hardware-rooted data security into the software stack itself allows the creation of a single, strong, contiguous data perimeter that extends elastically across all data states, cloud vendors and geographies. Much like virtualization, this enables most applications to run within a data perimeter with no code or IT-processes modifications. This will vastly reduce complexity, overall management and maintenance overhead. Now, data can realistically be protected end-to-end with a single, strong and consistent mechanism in what is effectively a confidential cloud.
Other elements of protecting data – including cloud-based installations – remain important, but this effectively makes data self-securing. This protects company secrets, intellectual property, machine-learning algorithms and data governed by privacy law and other regulations – by default. The closing of a simple memory-in-use vulnerability paves a clear way to decrease the complexity of current security, closes dangerous encryption gaps and brings critical advantage to organizations in the ongoing war against attackers.
Ayal Yogev CEO and Co-Founder, Anjuna Security
