What is secure computing?

The Secure Computing Alliance of the Linux Foundation defines secure computing as:

Confidential Computing protects data in use by performing computation in a hardware-based Trusted Execution Environment.

Confidential Computing Consortium

The following points are emphasized in this definition:

1. Secure computing protects data security during computing; 2. Secure computing requires the power of hardware.

Here is an illustration of these two points: in the cloud computing scenario, we can simplify cloud computing into three parts: data transmission, data operation and data storage.

The completeness of this three-part security solution varies. In the link of data transmission, the industry has very complete security standards and implementation, such as SSL, TLS. In the data storage link, cryptography also provides a very good solution, we can encrypt the data in a proper way after preservation, to prevent leaks in the storage link. In the calculation of data, there is no complete solution as in the other two. Secure computing aims to solve this problem.

How is secure computing implemented?

Let’s take Intel SGX technology as an example to see the specific technical solution.

Intel’s SGX technology builds secure computing environments (enclaves) within applications using the CPU as a trusted starting point for computing. Since the invention of computers, we have assumed that the CPU will do exactly what the software tells it to do. We just haven’t emphasized this. In today’s software development, all kinds of software work together on a hardware, the whole ecosystem is more and more complex, malicious software also appeared. To prevent malware damage, the CPU isolates a separate enclave environment for applications that need to be protected. Applications outside the enclave can neither observe nor modify code and data in the enclave, thus ensuring data security in the enclave. The CPU protects the enclave so strongly that even highly privileged operating systems and virtualization management software cannot break through this protection. In fact, it is not only against software, but also against peripheral hardware providers (e.g., motherboard makers, memory providers).

Intel SGX is the most mature secure computing product, but it is not the only secure computing product. Other hardware makers such as AMD, ARM and Nvidia are all rolling out secure computing products. All of these products are integrated software and hardware solutions with the following characteristics:

After understanding the concept of secure computing, this section introduces some typical scenarios of secure computing:

With a secure computing environment, users can safely place their applications in a shared cloud computing environment, where data used in computing and results can be encrypted and transmitted. This can unify the infrastructure architecture and avoid complex hybrid cloud deployments.

Data transactions and data services on the cloud are also possible. The data owner and the algorithm provider can respectively provide data and algorithm to the secure computing platform to complete the calculation without worrying about the problem of confidentiality disclosure.

Secure computing can also lead to more data collaboration. The data of all parties can be fused in a secure environment, so that the data can produce greater value.

In edge computing scenarios, compute nodes are deployed in a very complex environment and the machines are not controlled. Secure computing can effectively protect user data and privacy.

With so many application scenarios for secure computing, why haven’t we seen large-scale deployment? That’s because secure computing currently has a very big downside: ease of use. The specific performance is as follows:

** Application segmentation difficulty: ** It is difficult to transform an existing application into a secure computing application. You need to do code splitting.

Difficult deployment in ** scenarios: ** Secure computing is hardware dependent. The application scheduling system needs to be reformed in actual deployment.

** Security analysis difficult: ** An application using secure computing is not necessarily secure? The answer is no. This requires a very careful security analysis of the entire application.

Engineers from Ant group and Alibaba Group have come up with unique solutions to these problems. The first is to solve the difficult problem of application segmentation.

The open source Occlum project of Ant Group developed LibOS adaptation layer in the enclave, so that Linux applications can run in SGX environment without modification, completely solving the problem of application segmentation. Occlum was developed in Rust to ensure memory security. Supports multiple processes and encrypted file systems, and applications do not need to be modified. For example, applications developed based on SOFABoot, the financial level cloud native framework of Ant Group, can run in Occlum completely without modification.

👇 website link 🔗 : github.com/occlum/occl…

To solve the problem of difficult deployment, Ali Cloud launched Inclavare open source project.

Inclavare provides users with a secure computing container based on Occlum. Users can simply focus on the application itself, and Inclavare schedules computations to the appropriate compute nodes. Ant’s MORSE Multi-party security computing engine and MYTF blockchain computing platform provide solutions for different computing scenarios. Users no longer have to bear the high cost of security analysis.

Ant Group continues to invest in the field of secure computing, using the power of science and technology to protect data security, protect user privacy, and provide users with more secure services. Ant Group open-source TEE Security LibOS Occlum. Users can visit github.com/occlum/occl… Find all the implementation code. Users can review the source code in Occlum to ensure the security of the overall solution. You can also refer to the existing demo to learn how to use Occlum, and quickly get started with safe calculation.

Recommended Reading of the Week

  • Exploration and Practice of ant Cloud native Application runtime – ArchSummit Shanghai

  • Introduction to cloud native technology: Exploration and Practice of cloud native open Operation and Maintenance System

  • Improved Stability: New features for SOFARegistry V6

  • Finance-level capability becomes the core competitiveness, and service grid drives enterprise innovation

More articles please scan code to pay attention to “financial level distributed architecture” public number