WO2023041025A1

WO2023041025A1 - Cloud-technology-based computing node and cloud-technology-based instance management method

Info

Publication number: WO2023041025A1
Application number: PCT/CN2022/119266
Authority: WO
Inventors: 龚磊
Original assignee: 华为云计算技术有限公司
Priority date: 2021-09-18
Filing date: 2022-09-16
Publication date: 2023-03-23
Also published as: CN115828249A

Abstract

Disclosed in the embodiments of the present application are a cloud-technology-based computing node and a cloud-technology-based instance management method, which are used for improving the security, confidentiality and integrity of user programs and data and reducing the impact on the service performance. The cloud-technology-based computing node specifically comprises a host machine and an offloading card, wherein the offloading card is inserted into the host machine and establishes a communication channel with the host machine. An enclave instance is run on the host machine, the offloading card is provided with a trusted execution module that is allocated for the enclave instance, and an application program is provided in the enclave instance, wherein the enclave instance is used for sending, by means of the communication channel and to the trusted execution module, a confidential computation task that is generated by the application program; and the trusted execution module is used for processing the confidential computation task, and for returning a processing result to the enclave instance by means of the communication channel.

Description

Computing node based on cloud technology and instance management method based on cloud technology

This application claims the priority of the Chinese patent application with the application number 202111112194.0 and the title of the invention "Computing node based on cloud technology and instance management method based on cloud technology" submitted to the China Patent Office on September 18, 2021, the entire content of which Incorporated in this application by reference.

technical field

The embodiments of the present application relate to the field of cloud technology, and in particular, to a cloud technology-based computing node and a cloud technology-based instance management method.

Background technique

Cloud technology has been widely used in various fields, such as cloud services, and the security of the data environment of cloud services is related to the security of data privacy in the cloud.

Data in cloud services are generally divided into three forms: static data, data in transmission, and data in use. The former two can be encrypted to ensure data security. For data in use, confidential computing is generally used at present. Protecting the security of data in use mainly lies in providing a trusted execution environment (TEE), which is a secure area inside the central processing unit (CPU). The confidential data and software to be processed are placed in the TEE. The confidential data placed in the TEE and the software used to process the confidential data can only be accessed in an authorized manner, ensuring the confidentiality and integrity of the confidential data and reducing processing time. Confidential data is the attack surface of the software. The existing TEE is constructed based on a virtual machine monitor (virtual machine monitor, VMM). Specifically, part of the CPU and memory is isolated from the host computer through the VMM, and the isolated CPU and memory are used to create a TEE.

However, when the host machine is attacked maliciously, for example, after the VMM of the host machine is compromised, the CPU and memory in the host machine can be accessed through the VMM, the security of the TEE is affected, and the security of the TEE is reduced. In addition, the TEE will occupy the host machine. The resources on the host machine are affected.

Contents of the invention

The embodiment of the present application provides a computing node based on cloud technology, which is used to improve the security, confidentiality and integrity of user programs and data and reduce the impact on business performance. The embodiment of the present application also provides a corresponding instance management method based on cloud technology.

The first aspect of the present application provides a computing node based on cloud technology, including a host computer and an offload card, the offload card is inserted into the host computer and establishes a communication channel with the host computer, an enclave instance runs on the host computer, and the offload card A trusted execution module assigned to the Enclave instance is set on the Enclave instance, and an application program is set in the Enclave instance, wherein the Enclave instance is used to send the confidential computing tasks generated by the application program to the Trusted Execution Module through a communication channel; the Trusted Execution Module It is used to process confidential computing tasks and return the processing results to the Enclave instance through the communication channel.

The computing node based on cloud technology in this application can be a server. The server includes a host machine and an offloading card. One or more virtual machines (virtual machine, VM) can run on the host machine, and the virtual machine manager (Hypervisor) Execution, the virtual machine manager is also called virtual machine monitor (virtual machine monitor, VMM). The offload card is a standard external device of the host machine, also known as a smart card, which is inserted into the host machine and establishes a communication channel with the host machine. Enclave instance It is a trusted execution environment and a trusted isolation space.

The unloading card in this application is provided with a trusted execution module assigned to the Enclave instance. The trusted execution module is used to perform confidential calculations. The Enclave instance is provided with an application program. During the application process, the Enclave instance passes the application program through the communication channel. The generated confidential computing tasks are sent to the trusted execution module, and the trusted execution module can process the confidential computing tasks and return the processing results to the Enclave instance, thus completing the confidential computing tasks.

In the first aspect, the computing node based on cloud technology includes a host computer and an offload card. The offload card is inserted into the host computer and establishes a communication channel with the host computer. An enclave instance runs on the host computer. The trusted execution module assigned by the Enclave instance, where the Enclave instance is used to send confidential computing tasks to the trusted execution module; the trusted execution module is used to process confidential computing tasks, and the trusted execution module set on the uninstall card makes the TEE The trusted execution module responsible for confidential computing is physically isolated from the main instance through the host computer and the offloading card. After the host computer is attacked, it will not affect the confidential computing of the trusted execution module in the offloading card, thereby improving the security of user programs and data. , confidentiality and integrity. In addition, the trusted execution module in the offload card will not occupy the resources of the host computer, thereby reducing the impact of confidential computing on business performance.

In a possible implementation manner of the first aspect, a master instance is also running on the host machine, a virtual channel is established between the master instance and the Enclave instance, and the Enclave instance is also used to exchange information with the master instance through the virtual channel.

In this possible implementation, the master instance is also running on the host machine, and a virtual channel is established between the master instance and the enclave instance. The enclave instance can exchange information with the master instance through the virtual channel, such as notifying the master instance of the processing result. The virtual channel It can be a vsock channel, which can assist other security encryption schemes, such as digital signature algorithms such as RSA public key encryption, so that the master instance and the enclave instance can communicate securely and minimize the attack surface.

In a possible implementation of the first aspect, the unloading card is connected to the cloud management platform network, and the unloading card is also used to receive the master instance creation request sent by the cloud management platform, and notify the host computer to create the master instance through the communication channel.

In this possible implementation, the offload card is connected to the cloud management platform network, that is, the computing node based on cloud technology can be applied in the cloud data center, and the user can input a master instance creation request on the client side and send it to the cloud management platform through the Internet , the cloud management platform sends the master instance creation request to the offload card of the computing node through the internal network of the data center, and the offload card notifies the host to create the master instance through the communication channel, which improves the feasibility of the solution.

In a possible implementation manner of the first aspect, the master instance is further configured to notify the host machine to create an enclave instance based on an enclave instance creation command entered by a tenant logging into the master instance.

In this possible implementation, the computing node can be shared by multiple tenants. After logging in to the master instance, the tenant can enter the enclave instance creation command to notify the host to create an enclave instance, which improves the feasibility of the solution.

In a possible implementation manner of the first aspect, the number of Enclave instances is one.

In this possible implementation, one master instance is configured with one enclave instance, which improves the feasibility of the solution.

In a possible implementation of the first aspect, the number of Enclave instances is N, where N is a positive integer greater than or equal to 2, and the N Enclave instances handle different types of confidential computing tasks respectively.

In this possible implementation, a master instance can be configured with multiple enclave instances, and each enclave instance handles different types of confidential computing tasks. When the master instance starts an enclave instance, the application running on the encalve instance will generate a corresponding If a specific type of confidential computing task needs to be processed, the master instance only needs to start the corresponding Encalve instance, thereby improving the ability to split confidential computing tasks at a fine-grained level and ensuring the minimum authority.

In a possible implementation manner of the first aspect, the number of trusted execution modules is N, and each trusted execution module is connected to a different enclave instance.

In this possible implementation, the number of trusted execution modules is the same as the number of enclave instances, and each trusted execution module is connected to different enclave instances, which improves the feasibility of the solution.

In a possible implementation manner of the first aspect, one master instance corresponds to one trusted execution module.

In this possible implementation, the number of master instances is the same as the number of trusted execution modules, and the trusted execution modules corresponding to one master instance can be connected to different enclave instances to provide services, which improves the feasibility of the solution sex.

In a possible implementation manner of the first aspect, the communication channel includes a high-speed serial computer expansion bus PCIe channel and a computing express link CXL channel.

In this possible implementation, the communication channel can be a high-speed serial computer expansion bus (peripheral component interconnect express, PCIe) channel, or a computing express link (compute express link, CXL) channel, thereby improving the host computer and Offload the communication speed between cards.

The second aspect of the present application provides an instance management method based on cloud technology, which is applied to a computing node. The computing node includes a host computer and an offloading card. The offloading card is inserted into the host computer and establishes a communication channel with the host computer. The host computer An enclave instance runs on the enclave, the offloading card is provided with a trusted execution module assigned to the enclave instance, and the enclave instance is provided with an application program, the method includes the following steps: the enclave instance uses the communication channel to generate the confidential computing The task is sent to the trusted execution module; the trusted execution module processes the confidential computing task, and returns the processing result to the Enclave instance through the communication channel.

In the second aspect, the instance management method based on cloud technology is applied to the computing node. The computing node includes a host computer and an offloading card. The offloading card is inserted into the host computer and establishes a communication channel with the host computer. An enclave enclave runs on the host computer. For instance, the trusted execution module assigned to the Enclave instance is set on the uninstall card, the method includes: the Enclave instance sends the confidential computing task to the trusted execution module; the trusted execution module processes the confidential computing task, and the trusted execution module set on the uninstall card The execution module makes the trusted execution module responsible for confidential computing in the TEE physically isolated from the main instance through the host computer and the offloading card. After the host computer is attacked, it will not affect the confidential computing of the trusted execution module in the offloading card, thereby improving user The security, confidentiality and integrity of programs and data. In addition, the trusted execution module in the offload card will not occupy the resources of the host computer, thereby reducing the impact of confidential computing on business performance.

In a possible implementation of the second aspect, a master instance is also running on the host computer, and a virtual channel is established between the master instance and the Enclave instance, and the Enclave instance is used to exchange information with the master instance through the virtual channel. The method also includes : The Enclave instance notifies the master instance of the processing result through the virtual channel.

In a possible implementation of the second aspect, the unloading card is connected to the cloud management platform network, and the method further includes: the unloading card receives a master instance creation request sent by the cloud management platform, and notifies the host computer to create a master instance through a communication channel .

In this possible implementation, the unloading card is also connected to the network of the cloud management platform, that is, the instance management method based on cloud technology can be applied in the cloud data center, and the user can input the master instance creation request on the client and send it to the cloud through the Internet. The management platform, the cloud management platform sends the master instance creation request to the unloading card of the computing node through the internal network of the data center, and the unloading card notifies the host to create the master instance through the communication channel, which improves the feasibility of the solution.

In a possible implementation manner of the second aspect, the method further includes: the master instance notifies the uninstallation card to create the enclave instance and the trusted execution module based on the enclave instance creation command entered by the tenant who logs in to the master instance.

In this possible implementation, the computing node can be shared with multiple tenants. After logging in to the main instance, the tenant can enter the Enclave instance creation command to notify the uninstaller card to create an Enclave instance and a trusted execution module, which improves the feasibility of the solution. .

In a possible implementation manner of the second aspect, the number of Enclave instances is one.

In a possible implementation of the second aspect, the number of Enclave instances is N, where N is a positive integer greater than or equal to 2, and the N Enclave instances handle different types of confidential computing tasks respectively.

In a possible implementation manner of the second aspect, the number of trusted execution modules is N, and each trusted execution module is connected to a different enclave instance.

In a possible implementation manner of the second aspect, one master instance corresponds to one trusted execution module.

In a possible implementation manner of the second aspect, the communication channel includes a high-speed serial computer expansion bus PCIe channel and a computing express link CXL channel.

In this possible implementation manner, the communication channel may be a high-speed serial computer expansion bus channel, or a computing fast link channel, thereby increasing the communication speed between the host computer and the offloading card.

In the embodiment of this application, the computing node based on cloud technology includes a host computer and an offload card. The offload card is inserted into the host computer and establishes a communication channel with the host computer. An enclave instance runs on the host computer, and the offload card is set with The trusted execution module assigned to the Enclave instance, where the Enclave instance is used to send confidential computing tasks to the trusted execution module; the trusted execution module is used to process confidential computing tasks, and the trusted execution module set on the uninstall card makes the TEE The trusted execution module responsible for confidential computing is physically isolated from the main instance through the host computer and the offloading card. After the host computer is attacked, it will not affect the confidential computing of the trusted execution module in the offloading card, thereby improving the security of the TEE. In addition, The trusted execution module in the offload card will not occupy the resources of the host computer, thereby reducing the impact of confidential computing on business performance.

Description of drawings

Figure 1 is an architecture diagram of a cloud data center;

FIG. 2 is a schematic diagram of software layers of an embodiment of a computing node based on cloud technology provided by an embodiment of the present application;

FIG. 3 is a schematic diagram of a hardware layer of an embodiment of a computing node based on cloud technology provided by an embodiment of the present application;

FIG. 4 is a schematic diagram of the hardware layer of another embodiment of a computing node based on cloud technology provided by the embodiment of the present application;

FIG. 5 is a schematic diagram of an embodiment of a cloud-based instance management method provided by an embodiment of the present application;

FIG. 6 is a schematic diagram of another embodiment of the cloud technology-based instance management method provided by the embodiment of the present application.

Detailed ways

Embodiments of the present application are described below in conjunction with the accompanying drawings. Apparently, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Those of ordinary skill in the art know that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

The terms "first", "second" and the like in the specification and claims of the present application and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific sequence or sequence. It is to be understood that the terms so used are interchangeable under appropriate circumstances such that the embodiments described herein can be practiced in sequences other than those illustrated or described herein. Furthermore, the terms "comprising" and "having", as well as any variations thereof, are intended to cover a non-exclusive inclusion, for example, a process, method, system, product or device comprising a sequence of steps or elements is not necessarily limited to the expressly listed instead, may include other steps or elements not explicitly listed or inherent to the process, method, product or apparatus.

The embodiment of the present application provides a computing node based on cloud technology, which is used to improve the security, confidentiality and integrity of user programs and data and reduce the impact on business performance. The embodiment of the present application also provides a corresponding instance management method based on cloud technology. Each will be described in detail below.

As shown in Figure 1, the cloud data center includes a cloud management platform, an internal network of the data center, and multiple servers. Each server is the same, such as server 1 and server 2. The cloud management platform completes information interaction, and the cloud management platform completes information interaction with the client through the Internet. The network card is installed on the offload card. The offload card is also called a smart card. It is a standard external device for the server. The server also includes a host machine. The unloading card is inserted into the host machine and establishes a communication channel with the host machine. One or more virtual machines (virtual machines, VMs) can run on the host machine, such as virtual machine 1 and virtual machine 2. The host machine can also be called a cloud server (elastic compute service, ECS) or an elastic instance, etc., and the host machine is executed through a virtual machine manager (Hypervisor), which is also called a virtual machine monitor (virtual machine monitor, VMM).

Specifically, the cloud management platform provides an access interface (such as an interface or API), and tenants can operate the remote access interface of the client to register a cloud account and password on the cloud management platform and log in to the cloud management platform. After the authentication is successful, the tenant can further select and purchase a virtual machine with specific specifications (processor, memory, disk) on the cloud management platform. After the payment is successful, the cloud management platform provides the remote login account password of the purchased virtual machine. The client can remotely log in to the virtual machine, install and run the tenant's application in the virtual machine. The client of the cloud management platform receives the control plane commands sent by the cloud management platform, creates virtual machines on the server according to the control plane control commands, and performs full life cycle management of virtual machines. Therefore, tenants can create and manage virtual machines in the cloud data center through the cloud management platform. , Log in and operate the virtual machine.

Data typically must be decrypted in memory before an application can process it. This makes the data vulnerable to in-memory processing. Confidential computing typically addresses this issue by utilizing a hardware-based trusted execution environment (TEE), a secure enclave inside the CPU. TEEs are secured using embedded encryption keys and embedded proof mechanisms to ensure keys are only accessible by authorized application code. If malware or other unauthorized code tries to access the key - or authorized code is hacked or altered in any way the TEE will deny access to the key and cancel the computation. This way, sensitive data can remain protected in memory until the application tells the TEE to decrypt it for processing. During the decryption process and throughout the computing process, the operating system or the VMM in the VM, other computing stack resources, and the cloud provider and its employees cannot see the data.

Confidential computing protects sensitive data in use. When used with data-at-rest encryption and in transit with exclusive control of the keys, confidential computing eliminates the need for sensitive or highly regulated data sets and application workloads from inflexible, expensive local internet technology (IT) The biggest barrier to infrastructure migration to a more flexible and modern public cloud platform.

Confidential computing protects intellectual property. Confidential computing isn't just for data protection. TEEs can also be used to protect proprietary business logic, analytics functions, machine learning algorithms, or entire applications.

Confidential computing enables secure collaboration with partners on new cloud solutions. For example, a company can combine its sensitive data with another company's proprietary computing to create new solutions without requiring either company to share any data or intellectual property it does not wish to share.

Confidential computing can take the worry out of choosing a cloud provider. Confidential computing allows companies to choose the cloud computing service that best meets their technical and business requirements without having to worry about storing and processing customer data, know-how and other sensitive assets. This also helps mitigate any other competition concerns if the cloud provider also offers competing business services.

Confidential computing protects data processed at the edge. Edge computing is a distributed computing framework that brings enterprise applications closer to data sources, such as internet of things (IoT) devices or local edge servers. When used as part of a distributed cloud model, confidential computing can be used to protect data and applications on edge nodes.

The following describes the computing node based on cloud technology provided by the embodiment of this application in combination with the above-mentioned cloud data center architecture and the concept of confidential computing.

As shown in Figure 2, an embodiment of the computing node based on cloud technology provided by the embodiment of the present application includes a host machine 100 and an offloading card 200. The offloading card 200 is inserted into the host machine 100 and establishes a communication channel with the host machine 100. An enclave instance 120 is running on the host 100, and a trusted execution module 210 assigned to the enclave instance 120 is also provided on the unloading card 200, and an application program is arranged in the enclave instance 120, wherein the number of the enclave instance 120 is 1 , the communication channel may be a high-speed serial computer expansion bus (peripheral component interconnect express, PCIe) channel, or a computing express link (compute express link, CXL) channel.

Further, the master instance 110 is also running on the host machine 100, and a virtual channel is established between the master instance 110 and the enclave instance 120, and the enclave instance 120 is also used to exchange information with the master instance 110 through the virtual channel, for example, to notify the master instance 110 of the processing result .

Specifically, the host computer 100 also includes a virtual machine manager 140 and an instance simulator front-end module 130. The instance simulated by the instance simulator front-end module 130 is an abstract concept, not limited to virtual machines, containers, and functions. The instance simulator front-end module 130 is used to simulate various chipsets of computer equipment, life cycle management of central processing unit (CPU) in virtual machine, virtual machine exit (VM Exit) and other logic processing strongly related to virtual machine business, instance simulation The front-end module 130 specifically includes an intelligent control module 131, an isolated communication module 132, and an enclave security module 133. The intelligent control module 131 is used to control channels related to the life cycle of the Enclave instance 120, including but not limited to PCIe devices, virtualized IO Processing protocol (virtual IO, VirtIO) equipment, etc., and is responsible for isolating independent CPU and memory from the main instance 110 for use by the Enclave instance 120. The isolated communication module 132 is used to connect the master instance 110 and the safe communication channel of the Enclave instance 120, has perfect security control, and is a closed-loop connection (end to end) at the same time, and will not spread the attack surface of the virtual machine manager 140 itself, is A secure connection channel. The enclave security module 133 is used to complete trusted execution environment command-related operations for the enclave instance 120, and is used to transfer security-related commands and messages that need to be carried out in the enclave instance 120. The enclave security module 133 supports both virtual devices and real devices. Physical devices, including physical functions (physical function, PF), virtual functions (virtual function, VF) and standard (single root I/O virtualization, SR-IOV) endpoint devices that do not support PCIe sharing to virtual machines . The example simulator front-end module 130 also includes other modules related to its use, which is not limited in this embodiment of the present application.

The offload card 200 also includes a control plane logic component 230 and an instance emulator backend module 220 . The management control plane logic component 230 includes multiple components, such as components related to the life cycle of the virtual machine, such as components Nova and libvirt (components used to manage virtualization platforms) of open source OpenStack (an open source cloud computing management platform project) , kubernetes (for managing containerized applications on multiple hosts in the cloud platform), and components related to system installation, upgrade, deployment, and operation and maintenance, modules in the instance simulator back-end module 220 and instance simulator front-end modules 130 corresponds to one-to-one, which will not be repeated in this embodiment of the present application.

More specifically, the intelligent control module 131 includes a life cycle management module 1311 , a memory isolation module 1312 , a CPU isolation module 1313 and a virtual device simulation module 1314 . The lifecycle management module 1311 is used to create, destroy and reconfigure the Enclave instance 120, and is also responsible for the connection management with the instance simulator backend module 220. The memory isolation module 1312 is used to perform memory isolation operations, which need to be connected with the virtual The machine manager 140 cooperates to ensure that the users of the main instance 110 or simulation programs have no access to the memory of the enclave instance 120. The CPU isolation module 1313 is used to perform CPU isolation operations and needs to work with the virtual machine manager 140 to ensure that the main instance The user or simulation program of 110 has no right to use the CPU independently used by the Enclave instance 120. The virtual device simulation module 1314 is used to simulate the device of the master instance 110 command channel, which can be a VirtIO device or any other peripheral component interconnection standard (peripheral component interconnect, PCI) devices, memory-mapped I/O (MMIO) analog devices, etc.

The isolated communication module 132 includes a control device module 1321 , an unowned communication management module 1322 and a rights management module 1323 . The control device module 1321 is mainly used to establish a secure internal channel connection between the master instance 110 and the Enclave instance 120. The channel can be a VirtIO device, a network card device, and a shared memory device. The function of the virtual machine manager 140 directly performs end-to-end communication between the main instance 110 and the enclave instance 120, which does not expand the attack surface of the virtual machine manager 140, and also ensures the absolute security of the enclave instance 120. The authority management module 1323 It is used to control and manage the permissions of the communication channel to ensure that no other abnormal users or escaped attackers can access the communication channel.

The enclave security module 133 includes a device management module 1331 , a backend communication management module 1332 , a resource management module 1333 and a security information encapsulation module 1334 . The device management module 1331 is used for the simulation of the lightweight trusted execution environment, and includes the device management function of the security module device, the virtual security module device or the direct access to the real module device. The back-end communication management module 1332 establishes a connection with the trusted execution module 210 of the unloading card 200, and transmits the request and response information of the Enclave instance 120. The resource management module 1333 is responsible for the management resource allocation or release of the entire Enclave instance 120, etc., and the security information encapsulation module 1334 securely encapsulates the request issued by the Enclave instance 120, the security information encapsulation module 1334 is optional, and is not required in some scenarios, such as the direct scenario.

The trusted execution module 210 includes a message management module 211 , a security processing module 212 and a hardware acceleration module 213 . The message management module 211 is used to perform a handshake connection with the instance emulator front-end module 130, including the establishment of a communication channel, automatic reconnection, and message distribution. The security processing module 212 is used to perform security operation processing related to confidential computing tasks, such as encryption and decryption, random number generation processing, and certificate generation logic. The hardware acceleration module 213 is used to enable the capability of the security chip or TEE, store some confidential information, such as keys, values of the platform configuration register (PCR), etc. in the hardware, or use hardware to accelerate security operations. The entire trusted execution module 220 can be placed entirely or partially in the TEE, thereby constructing a completely isolated confidential computing environment.

As shown in FIG. 3 , at the hardware level, the host machine 310 includes a CPU 311 , a memory 312 and a root multiplexer (root complex, RC) chip 313 . The RC chip 313 is a chip defined in the PCIe specification, and its function is to provide a hardware interface for the CPU 311 of the host machine 310 to send and receive PCIe messages to and from external devices. The host machine 310 and the offloading card 320 are interconnected using the standard PCIe protocol, that is, the communication channel is a PCIe channel, and the communication channel can also be a CXL channel, and the RC chip 313 can be replaced at this time.

The uninstall card 320 includes a CPU 321 , a memory 322 , an endpoint (end point, EP) chip 323 and a security chip 324 . Wherein the EP chip 324 is a chip defined in the PCIe specification, and its function is to send and receive PCIe messages to the host computer 310 as a peripheral interface, that is, the communication channel is a PCIe channel, and the communication channel can also be a CXL channel. At this time, the EP chip 323 Substitutions can be made. In addition, the security chip 324 can be a TEE security chip, and the TEE security chip can also be set in the CPU 321 , so as to further strengthen the security of confidential computing through hardware.

During application, the application program (application, APP) running on the Enclave instance generates confidential computing tasks. The application program is self-developed by the user, that is, the user pre-creates the application program according to his own confidential computing needs. The specific confidential computing task can be obtained Random numbers or obtain key information, etc., and then the Enclave instance sends the confidential computing task to the trusted execution module. After computing the task, parse the header information used in the management of the confidential computing task without parsing the content of the specific confidential computing task. The enclave security module continues to call the back-end communication management module to send the assembled confidential computing task request through a secure communication channel. To uninstall the trusted execution module of the card, after the trusted execution module receives the request of the confidential computing task, it analyzes and processes it through the security processing module, and the trusted execution module can call the corresponding logic of the hardware acceleration module according to the category of the confidential computing task. To speed up the processing, the trusted execution module returns the processing result to the Enclave instance after obtaining the processing result, specifically, it returns to the enclave security module of the front-end module of the instance simulator first, and the enclave security module receives the processing result and sends the processing result to the Enclave instance, and finally the Enclave instance can also notify the master instance of the processing result through the virtual channel to complete confidential computing tasks.

Furthermore, if parameters need to be obtained from the outside when processing confidential computing tasks, the Enclave instance can communicate with the master instance through a virtual channel, and use the master instance as a proxy to obtain these parameters from the external network.

In addition, when the computing node fails, the host and the smart card can be migrated to other computing nodes together to ensure the portability of cloud services and the elasticity of virtualization, and the bare metal server will not be consumed during the hot migration process Server resources, such as CPU and memory, do not affect tenant business performance, such as performance jitter.

In the embodiment of this application, the computing node based on cloud technology includes a host computer and an offload card. The offload card is inserted into the host computer and establishes a communication channel with the host computer. An enclave instance runs on the host computer, and the offload card is set with The trusted execution module assigned to the Enclave instance, where the Enclave instance is used to send confidential computing tasks to the trusted execution module; the trusted execution module is used to process confidential computing tasks, and the trusted execution module set on the uninstall card makes the TEE The trusted execution module responsible for confidential computing is physically isolated from the main instance through the host computer and the offloading card. After the host computer is attacked, it will not affect the confidential computing of the trusted execution module in the offloading card, thereby improving the security of user programs and data. In addition, the trusted execution module in the offload card will not occupy the resources of the host computer, thereby reducing the impact of confidential computing on business performance.

As shown in FIG. 4 , in another embodiment of the computing node based on cloud technology provided by the embodiment of the present application, the offload card 420 is also connected to the cloud management platform 404 network, that is, the computing node 400 is applied to the data center 400 .

For the hardware level, the host machine 410 includes a CPU 411 , a memory 412 and a root multiplexer (root complex, RC) chip 413 . The RC chip 413 is a chip defined in the PCIe specification, and its function is to provide a hardware interface for the CPU 411 of the host machine 410 to send and receive PCIe messages to and from external devices. The host computer 410 and the offloading card 420 are interconnected using the standard PCIe protocol, that is, the communication channel is a PCIe channel, and the communication channel can also be a CXL channel, and the RC chip 413 can be replaced at this time. The uninstall card 420 includes a CPU 421 , a memory 422 , an endpoint (end point, EP) chip 423 and a security chip 424 . Wherein the EP chip 424 is a chip defined in the PCIe specification, and its function is to send and receive PCIe messages to the host computer 410 as a peripheral interface, that is, the communication channel is a PCIe channel, and the communication channel can also be a CXL channel. At this time, the EP chip 423 Substitutions can be made. In addition, the security chip 424 can be a TEE security chip, and the TEE security chip can also be set in the CPU 421 , so as to further strengthen the security of confidential computing through hardware.

In addition, the offloading card 420 also includes a network card 425, and the offloading card 420 is connected to global storage resources and network resources through the network card 425. Specifically, the network card 425 can be connected to the cloud management platform 404 through the data center internal network 403, the data center internal network 403 can also be connected to the network cards of other computing nodes 402, and the cloud management platform 404 is connected to the client 406 through the Internet 405. The cloud management platform client in the embodiment is offloaded in the network card 425 of the offloading card 420 , that is, external users do not directly communicate with the host machine 410 , but communicate securely with the host machine 410 through the offloading card 420 .

Before performing confidential computing tasks, the user needs to create a master instance and an enclave instance, and the offload card receives the master instance creation request sent by the cloud management platform, and notifies the host to create a master instance through the communication channel. The master instance notifies the host to create an enclave instance based on the enclave instance creation command entered by the tenant who logs in to the master instance, and the master instance also notifies the uninstaller card to create a trusted execution module based on the enclave instance creation command.

Specifically, the administrator or user can send a master instance creation request. Because the cloud management platform client is uninstalled on the uninstall card, the management component of the uninstall card will first receive the master instance creation request sent by the cloud management platform, and then assemble the corresponding master instance. The configuration file of the instance, which includes the CPU type, number of CPUs, memory size, disk, network card, mouse, etc. After the configuration file is assembled, the uninstallation card notifies the host to create a master instance through the communication channel, that is, the configuration file and the master instance The creation request is sent to the host, and the host will start the front-end module of the instance simulator according to the content of the configuration file, and call the virtual device simulation module of the front-end module of the instance simulator to work. Uninstalling the card after startup will also start the back-end module of the instance simulator. And establish a connection channel with the front-end module of the instance emulator for the transmission of subsequent messages, thereby completing the creation of the master instance, and the master instance starts working.

After the user purchases the master instance, the user becomes a tenant. The tenant can log in to the master instance, select Enclave on the interface of the master instance or during the creation process of the master instance, and then initiate an Enclave instance creation command on the master instance through the Enclave tool. The enclave instance creation command includes Information such as the number of CPUs and memory size required by the enclave instance, the life cycle management module of the intelligent command control module of the instance emulator front-end module of the main instance receives the enclave instance creation command, and will analyze the parameters of the enclave instance creation command, and prepare the enclave instance creation command before starting Prepare, and then the intelligent control module calls the memory isolation module according to the parameters, and cooperates with the virtual machine manager to isolate the memory of the corresponding size from the main instance. The intelligent control module also calls the CPU isolation module according to the parameters, and the virtual machine manager. In cooperation, the corresponding CPU resources are allocated from the main instance for the exclusive use of the Enclave instance, and then the life cycle management module starts the Enclave instance of the host machine, and calls the device management module of the enclave security module to perform device simulation. Then the front-end module of the instance simulator of the main instance notifies the back-end module of the instance simulator of the unloading card through the communication channel to start building a trusted execution environment. The back-end module of the instance simulator pulls up the trusted execution module and uses the message management module to communicate with the host The enclave instance establishes a secure connection. At this time, the enclave instance and the trusted execution module start to work, and can perform subsequent confidential computing tasks.

Further, the number of Enclave instances is N, and N is a positive integer greater than or equal to 2. N Enclave instances process different types of confidential computing tasks respectively, and the number of trusted execution modules on the offload card is also N. And each trusted execution module is respectively connected with different enclave instances, that is, each enclave instance is connected with a trusted execution module. When executing a confidential computing task, when the master instance starts an Encalve instance, the application program running on the Encalve instance will generate a corresponding confidential computing task. If a specific type of confidential computing task needs to be processed, the master instance only needs to start the corresponding Encalve instance. Can. In addition, a master instance can also correspond to a trusted execution module, and the number of master instances and trusted execution modules is the same, and the trusted execution modules corresponding to a master instance can be connected to different enclave instances to provide services.

In the embodiment of the present application, the computing node is applied to the cloud data center, and the unloading card is also connected to the cloud management platform network, thereby creating a master instance and an enclave instance, which improves the feasibility of the solution. In addition, the number of enclave instances is multiple. The ability to fine-grained split confidential computing tasks has been improved to ensure that permissions are minimized.

As shown in Figure 5, in an embodiment of the instance management method based on cloud technology provided by the embodiment of the present application, the method includes:

501. The Enclave instance sends the confidential computing task generated by the application program to the trusted execution module through a communication channel.

502. The trusted execution module processes the confidential computing task, and returns the processing result to the Enclave instance through the communication channel.

The instance management method based on cloud technology is applied to computing nodes. Computing nodes include a host and an offload card. The offload card is inserted into the host and establishes a communication channel with the host. Enclave instances run on the host. The card is also provided with a trusted execution module assigned to the Enclave instance, and the Enclave instance is provided with an application program, wherein the number of the Enclave instance is 1, and the communication channel includes a high-speed serial computer expansion bus PCIe channel and a computing fast link CXL aisle.

The Enclave instance of the computing node sends the confidential computing tasks generated by the application to the trusted execution module through the communication channel. The trusted execution module of the computing node processes the confidential computing tasks and returns the processing results to the Enclave instance to complete the confidential computing tasks. For the specific implementation of the instance management method based on cloud technology provided by the embodiment of the application, reference may be made to the foregoing description of computing nodes based on cloud technology, and details will not be repeated in this embodiment of the application.

In the embodiment of this application, the instance management method based on cloud technology is applied to the computing node. The computing node includes a host computer and an offload card. The offload card is inserted into the host computer and establishes a communication channel with the host computer. An enclave runs on the host computer. For an enclave instance, a trusted execution module assigned to the enclave instance is set on the unloading card. The method includes: the enclave instance sends the confidential computing task to the trusted execution module; The trusted execution module makes the trusted execution module responsible for confidential computing in the TEE physically isolated from the main instance through the host computer and the offloading card. After the host computer is attacked, it will not affect the confidential computing of the trusted execution module in the offloading card, thereby improving The security, confidentiality and integrity of user programs and data. In addition, the trusted execution module in the offload card will not occupy the resources of the host computer, thereby reducing the impact of confidential computing on business performance.

As shown in Figure 6, in another embodiment of the instance management method based on cloud technology provided by the embodiment of the present application, the method includes:

601. The uninstallation card receives a master instance creation request sent by the cloud management platform, and notifies the host computer to create the master instance through a communication channel.

602. The master instance notifies the unloading card to create an enclave instance based on the enclave instance creation command input by the tenant who logs in to the master instance.

603. Based on the Enclave instance creation command, the master instance notifies the offload card to create a trusted execution module.

604. The Enclave instance sends the confidential computing task generated by the application program to the trusted execution module through the communication channel.

605. The trusted execution module processes the confidential computing task, and returns the processing result to the Enclave instance.

606. The enclave instance notifies the master instance of the processing result through the virtual channel.

The instance management method based on cloud technology is applied to computing nodes. Computing nodes include a host computer and an offload card. The offload card is inserted into the host computer and establishes a communication channel with the host computer. Enclave instances run on the host computer. The offload card There is also a trusted execution module assigned to the Enclave instance, and an application program is set in the Enclave instance, a master instance runs on the host computer, a virtual channel is established between the master instance and the Enclave instance, and the Enclave instance is used to communicate with the master instance through the virtual channel. The instance performs information exchange, and the offload card is connected to the network of the cloud management platform. The number of Enclave instances is N, and N is a positive integer greater than or equal to 2. N Enclave instances handle different types of confidential computing tasks respectively. The trusted execution module The number is N, and each trusted execution module is connected to a different Enclave instance, wherein the communication channel includes a high-speed serial computer expansion bus PCIe channel and a computing express link CXL channel. In addition, a master instance can also correspond to a trusted execution module, and the number of master instances and trusted execution modules is the same, and the trusted execution modules corresponding to a master instance can be connected to different enclave instances to provide services.

The unloading card of the computing node receives the master instance creation request sent by the cloud management platform, and notifies the host to create a master instance through the communication channel, and then the master instance of the computing node notifies the unloading card to create an enclave based on the enclave instance creation command entered by the tenant who logged in to the master instance instance, and based on the Enclave instance creation command, the offloading card is notified to create a trusted execution module. If the main instance of the computer node needs to perform some confidential computing task, it only needs to start the Enclave instance corresponding to the confidential computing task, and the Enclave instance of the computing node The confidential computing tasks generated by the applications running on the enclave instance are sent to the trusted execution module through the communication channel. The trusted execution module of the computing node processes the confidential computing tasks and returns the processing results to the enclave instance. Finally, the enclave instance of the computing node The master instance is notified of the processing result through the virtual channel to complete the confidential computing task. For the specific implementation of the instance management method based on cloud technology provided by the embodiment of this application, please refer to the foregoing description about computing nodes based on cloud technology. repeat.

In the embodiment of this application, the instance management method based on cloud technology is applied to the computing node, and the computing node is applied to the cloud data center, and the offloading card is also connected to the cloud management platform network, thereby creating a master instance and an enclave instance, which improves the reliability of the solution. In addition, the number of enclave instances is multiple, which improves the ability to split confidential computing tasks at a fine-grained level and ensures the minimum authority.

Those skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the method described above can refer to the corresponding process in the foregoing unit embodiments, which will not be repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or May be integrated into another system, or some features may be ignored, or not implemented. In another point, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, each unit may exist separately physically, or two or more units may be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

If the integrated unit is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or part of the contribution to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , including several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, read-only memory), random access memory (RAM, random access memory), magnetic disk or optical disc, etc., which can store program codes. .

Claims

A computing node based on cloud technology, characterized in that it includes a host machine and an offloading card, the offloading card is inserted into the host machine and establishes a communication channel with the host machine, the host machine runs a flying An Enclave instance, the unloading card is provided with a trusted execution module allocated for the Enclave instance, and the Enclave instance is provided with an application program, wherein,

The Enclave instance is configured to send the confidential computing task generated by the application program to the trusted execution module through the communication channel;

The trusted execution module is configured to process the confidential computing task, and return the processing result to the Enclave instance through the communication channel.
The computing node according to claim 1, wherein a master instance is also running on the host computer, and a virtual channel is established between the master instance and the Enclave instance, wherein:

The Enclave instance is also used to exchange information with the master instance through the virtual channel.
The computing node according to claim 2, wherein the offloading card is connected to the cloud management platform network, wherein:

The unloading card is configured to receive the master instance creation request sent by the cloud management platform, and notify the host computer to create the master instance through the communication channel.
The computing node according to claim 2, wherein,

The master instance is further configured to notify the host machine to create the enclave instance based on an enclave instance creation command entered by a tenant logging into the master instance.
The computing node according to claim 4, wherein,

The master instance is further configured to notify the unloading card to create the trusted execution module based on the enclave instance creation command.
The computing node according to any one of claims 1 to 5, wherein the number of Enclave instances is one.
The computing node according to any one of claims 2 to 5, wherein the number of Enclave instances is N, and N is a positive integer greater than or equal to 2, and the N Enclave instances process different types of Confidential computing tasks.
The computing node according to claim 7, wherein the number of the trusted execution modules is N, and each of the trusted execution modules is respectively connected to different Enclave instances through the communication channel.
The computing node according to any one of claims 1 to 8, wherein the communication channel includes a high-speed serial computer expansion bus PCIe channel and a CXL channel.
A cloud-based instance management method, characterized in that the method is applied to a computing node, and the computing node includes a host machine and an offloading card, and the offloading card is inserted into the host machine and connected to the host machine A communication channel is established, an Enclave instance of an enclave runs on the host machine, a trusted execution module allocated for the Enclave instance is set on the offloading card, and an application program is set in the Enclave instance, and the method Include the following steps:

The Enclave instance sends the confidential computing task generated by the application program to the trusted execution module through the communication channel;

The trusted execution module processes the confidential computing task, and returns a processing result to the Enclave instance through the communication channel.
The method according to claim 10, wherein a master instance is also running on the host computer, a virtual channel is established between the master instance and the Enclave instance, and the Enclave instance is used to communicate with the Enclave instance through the virtual channel The main instance performs information interaction, and the method also includes:

The Enclave instance notifies the master instance of the processing result through the virtual channel.
The method according to claim 11, wherein the unloading card is connected to a cloud management platform network, and the method also includes:

The unloading card receives the master instance creation request sent by the cloud management platform, and notifies the host machine to create the master instance through the communication channel.
The method according to claim 11, characterized in that the method further comprises:

The master instance notifies the host machine to create the enclave instance based on an enclave instance creation command entered by a tenant logging into the master instance.
The method according to claim 13, further comprising:

The master instance notifies the unloading card to create the trusted execution module based on the enclave instance creation command.
The method according to any one of claims 10 to 14, wherein the number of Enclave instances is one.
The method according to any one of claims 11 to 14, wherein the number of Enclave instances is N, and N is a positive integer greater than or equal to 2, and the N Enclave instances process different types of secrets respectively computing tasks.
The method according to claim 16, wherein the number of the trusted execution modules is N, and each of the trusted execution modules is respectively connected to different Enclave instances through the communication channel.
The method according to any one of claims 10 to 17, wherein the communication channel comprises a high-speed serial computer expansion bus PCIe channel and a CXL channel.