Azure Confidential Computing and Zero Trust
by Razi Rais, Jeff Birnbaum, Graham Bury, Vikas Bhatia
Appendix B. Confidential Computing and Other Privacy-Enhancing Technologies
The main objective of privacy-enhancing technologies (PETs) is to facilitate computation on data without revealing the data to the party conducting the computation or to the hosting platform. This approach to privacy is fundamentally distinct from traditional cryptographic primitives, which typically focus on data confidentiality but do not allow any operations on encrypted data (ciphertext). However, in recent years, the demand for production-ready PETs has increased tremendously. This was initially driven by the massive shift toward cloud computing in the 2010s, especially the move to the public cloud, and was recently accelerated by the emergence of the much anticipated AI disruption, driven by technologies such as generative AI, which requires large language models to be trained in the cloud and possibly even on the edge.
Over the years, a variety of PETs with different degrees of maturity have emerged, including but not limited to homomorphic encryption, trusted execution environments, Trusted Platform Modules, and more. Although exhaustive coverage is beyond the scope of this report, Table B-1 compares a few PETs. Also, please keep in mind that, depending on the use case, these technologies may be complementary rather than mutually exclusive.
| Confidential computing/hardware TEEa | Homomorphic encryptionb | Trusted Platform Modulec |
|---|---|---|
| Data integrity ... |
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access