Confidential Computing: A Critical Step Towards Zero Trust

Evidently, traditional methods of cybersecurity are not matching the capabilities of today’s malicious actors in cyberspace. Cybersecurity has long since ceased to be an issue affecting ‘others’ and is now relevant to anyone with an internet connection, be it an iPhone user, a hospital or a vaccine manufacturer. Any internet user can now check their passwords on IOS or Google and will undoubtedly be confronted by a list of their frequently used apps that have been subject to a security breach and advised to change their credentials.
Likewise, the recent ‘solar winds attack’ invaded significant parts of the US government machinery, including the defense establishment, and remained hidden, gathering intelligence for many months.

A zero trust strategy seeks to eliminate the most common entry methods for malicious behaviour by accepting that errors in coding and architecture are only part of the problem. Human fallibility is just as great. Whether by illegal means or sophisticated phishing, obtaining access to secure systems appears to be becoming increasingly easy for those determined and motivated.

Imagine going to a café by bicycle. When you arrive, you lock the bike with two different locks, activate an alarm, and then go inside for coffee. In a zero-trust world, you don’t bother with the locks and alarms. You always sit outside with the bicycle next to you. Perversely, the unlocked bike will be better protected.

Confidential Computing aims to give that same level of security and acknowledges that organizations are no longer in a position to move freely within their own space. Proprietary data centers are subject to breaches by an insider threat, whether criminal or inadvertent as well as external players. Public clouds are equally vulnerable.

Compliance with common standards can neither be assured nor even verified against sophisticated attacks.

Perimeters that seek to protect the internal or intranet business of an organization from external cyberspace are increasingly easy to breach, especially with web services serving so many clients simultaneously.

Constantly ‘improving’ systems by patching and building onto legacy architecture ultimately degrades the building and threatens it with collapse

To combat these issues, many different types of ‘locks and alarms’ have been added to the cybersecurity arsenal to plug the gaps. These include:

Fine-grained access control whereby each item of data has its own security and access protocol

Microsegmenting data: this is akin to watertight bulkheads on a ship. If one section is breached, it protects the next one preventing the whole ship from sinking.

Multi-factor authentication, requiring users to have access to multiple devices simultaneously, is perhaps the most familiar technique as it is offered to almost all mobile phone users

The above elements have become increasingly less effective. Newer technologies such as AI, IOT, 5G, and quantum computing exponentially add vastly more information and related processing and to many more physical locations.This requires new and more relevant approaches to zero trust.Confidential computing is a new, innovative and critical element to achieving the protection zero trust offers.

It does so by protecting and isolating data and applications while in use.

There are multiple approaches to CC each with their own pros and cons.

Secure enclaves embed segmentation into the hardware with additional bespoke software defenses.

Homomorphic encryption allows users to access necessary data within an enclave without decrypting it. In layman’s terms, what happens in Vegas, stays in Vegas.

iSGX is an Intel technology that allows the CPU to decrypt specific portions of memory ‘on the fly’ but uses code only available within the enclave itself. This method blocks any malicious code from reading or using the data, even if it’s accessed.

Truly Confidential computing should employ hardware secure enclaves across all operating systems, and neither require any code to be modified nor impact performance in speed or accuracy of processing.

Secure enclaves are not the only weapons in the arsenal.Confidemtial computing platforms should be underpinned by ultrasecure firewalls to monitor messages going in and out, a mini hardware security module for ultimate authentication and secure remote management. The platform must embed access and approval policies in their own enclaves and includes CPU’s and/or GPU’s for apps.The design of a platform should incorporate a policy of zero trust, even for administrators, from the outset.

The most user-friendly hardware for the platform could be a single tamper-proof box, customizable for clients, that immeasurably improves security from malicious intruders and effectively provides a secure data center, especially for apps and data on the edge. When an organization is confident that it has a watertight zero trust IT system, integration with a public cloud can be made with vastly increased confidence.

Such a powerful approach to a confidential computing platform provides an incredibly effective zero trust solution. Zero trust in a box.

System breaches undermine customer confidence, and ultimately can destroy an organization’s reputation in a moment. Bespoke solutions as outlined here not only use locks and alarms, they sit next to the bike 24/7.

David Hochhauser is an experienced executive in the Security industry, with a broad background and solid leadership skills in security, enterprise software and cloud-based services. He is CRO at Hub Security focusing on an innovative key management platform that effectively and efficiently protects highly sensitive information. He has been an executive at both large and small companies including IBM, CA Technologies and Eurekify, which was acquired, and with regional, North American and WW responsibilities David has a Master’s of Science in Physics from New York University.