Confidential Computing

real-cis builds on more than 15 years of experience in secure cloud platform development and operations.

While data can be well protected against unauthorized access during transmission from the users' end devices to the cloud or, in the Internet of Things, from the "things" to the cloud, as well as during permanent storage by cryptographic encoding, it is hard to protect data and application code during computing in the processor cores (CPU), the random access memories (RAM), and on system buses.

At these critical components data and code are accessible for attackers and privileged staff. Protection against unauthorized access by personnel of the cloud operator or external attackers is merely provided by regulations concerning organizational measures and monitoring with logs and records of the administration activities. 

In systems without Confidential Computing, it is often, despite the monitoring of administrators by Privileged Access Management systems, not possible to distinguish between maintenance activities of administrators with privileged access necessary for operations and unauthorized access to data being processed on behalf of the cloud users, .

Confidential computing measures can be used to provide additional technical barriers to protect against abuse by external or internal perpetrators of the operator's privilege, which necessarily had to be accepted prior to the availability of Confidential Computing technology. 

In addition, Confidential Computing can protect against attacks via exploits or backdoors that may be contained in the operating system or the various components of the middleware. Confidential computing can also protect against possible side-channel attacks on data as it is processed.
 

Implementations

The core functionality of Confidential Computing is to provide a secure and trusted execution environment (TEE). The perimeter of the TEE can be the CPU of the processor chip of the cloud servers with an associated memory "enclave" (chip-based processor TEE) or a chassis- or rack-level capsule equipped with penetration sensors encapsulating one or more cloud servers (penetration sensor-based multi-server TEE).

In these systems, in accordance with the standard for the highest protection class for hardware security modules, FIPS 140-2, a precautionary deletion of the cloud users' unencrypted data is carried out when the penetration sensors are triggered (data clean-up). Both types of implementation have in common, that

  • the data to be processed is only decrypted within the TEE and then re-encrypted before leaving the TEE, and,
  • the secure operating state, the so-called "sealed state" of the TEE, is reached only after an audit and initialization of the "enclave" or TEE .

The chip-based processor TEEs protect in particular against attacks through vulnerabilities in the operating systems and the system software components. The penetration sensor-based multi-server TEEs better protect against physical and side-channel attacks. The best protection is provided by the combinations of penetration-sensor-based multi-server TEE and chip-based processor TEE. 

For more information, do not hesitate to contact us: