Microsoft is committed to enabling the industry to move from ‘computing in the clear’ to ‘computing confidentially’. Why? Common scenarios confidential computing have enabled include:
- Multi-party rich and secure data analytics
- Confidential blockchain with secure key management
- Confidential inferencing with client and server measurements & verifications
- Microservices and secure data processing jobs
The public preview of confidential computing nodes powered by the Intel SGX DCsv2 SKU with Azure Kubernetes Service brings us one step closer by securing data of cloud native and container workloads. This release extends the data integrity, data confidentiality and code integrity protection of hardware-based isolated Trusted Execution Environments (TEE) to container applications.
Azure confidential computing, based on Intel SGX-enabled virtual machines, continues encrypting data while the CPU is processing it—that’s the “in use” part. This is achieved with a hardware-based TEE that provides a protected portion of the hardware’s processor and memory. Users can run software on top of the protected environment to shield portions of code and data from view or modification outside of the TEE.
Expanding Azure confidential computing deployments
Developers can choose different application architectures based on whether they prefer a model with a faster path to confidentiality or a model with more control. The confidential nodes on AKS support both architecture models and will orchestrate confidential application and standard container applications within the same AKS deployment. Also, developers can continue to leverage existing tooling and dev ops practices when designing highly secure end-to-end applications.
During our preview period, we have seen our customers choose different paths towards confidential computing:
- Most developers choose confidential containers by taking an existing unmodified docker container application written in a higher programming language like Python, Java etc. and chose a partner like Scone, Fortanix and Anjuna or Open Source Software (OSS) like Graphene or Occlum in order to “lift and shift” their existing application into a container backed by confidential computing infrastructure. Customers chose this option either because it provides a quicker path to confidentiality or because it provides the ability to achieve container IP protection through encryption and verification of identity in the enclave and client verification of the server thumbprint.
- Other developers choose the path that puts them in full control of the code in the enclave design by developing enclave aware containers with the Open Enclave SDK, Intel SGX SDK or chose a framework such as the Confidential Consortium Framework (CCF). AI/ML developers can also leverage Confidential Inferencing with ONNX to bring a pre-trained ML model and run it confidentially in a hardware isolated trusted execution environment on AKS.
One customer, Magnit, chose the first path. Magnit is one of the largest retail chains in the world and is using confidential containers to pilot a multi-party confidential data analysis solution through Aggregion’s digital marketing platform. The solution focuses on creating insights captured and computed through secured confidential computing to protect customer and partner data within their loyalty program.
We have aggregated more samples of real use cases and continue to expand this sample list here: https://aka.ms/accsamples.
How to get going
- “Lift and shift” your existing application by choosing a partner such as Anjuna, Fortanix, or Scone or by using Open Source Software such as Graphene or Occlum.
or - Build a new confidential application from scratch using an SDK such as the Open Enclave SDK.
Confidential computing, through its isolated execution environment, has broad potential across use cases and industries; and with the added improvements to the overall security posture of containers with its integration to AKS, we are excited and eager to learn more about what business problems you can solve.
Get started today by learning how to deploy confidential computing nodes via AKS.