Methods and systems for creating confidential cloud computing platforms in a split architecture
By employing an authentication and key management system in a split resource architecture, combined with a cloud scheduler and monitoring agent, and dynamically allocating processor and storage hardware resources, and utilizing TEE modules and security controllers, the challenge of protecting distributed components in a traditional cloud computing platform with a split resource architecture is solved, achieving efficient security and integrity protection for confidential virtual machines.
Patent Information
- Application Number
- CN202480090074.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-28
- Publication Date
- 2026-11-17
AI Technical Summary
Traditional cloud computing platforms struggle to effectively protect the confidentiality of distributed components in a split-resource architecture. Existing TEE technology is ill-suited for split-resource architectures, lacks scalability and a unified authentication method, and is unable to protect distributed components across split-resource architectures.
An authentication and key management system is adopted, combined with a cloud scheduler and monitoring agent, to dynamically allocate processor and storage hardware resources. The security of confidential virtual machines is ensured by using TEE modules and security controllers. The authentication and key management system provides end-to-end security policies and authentication processes to ensure data confidentiality and integrity.
It enables efficient and reliable protection of confidential virtual machines in a split resource architecture, ensuring data security and integrity, improving resource utilization, and enhancing user experience and platform performance.
Smart Images

Figure CN123125579A_ABST
Abstract
Description
Technical Field
[0001] This invention generally relates to the field of cloud computing, and more specifically, to methods and systems for creating a confidential cloud computing platform in a split resource architecture, the confidential cloud computing platform including at least one confidential VM among a plurality of confidential virtual machines (VMs). Background Technology
[0002] The ever-increasing demand for performance in cloud data centers is driving a fundamental transformation of cloud infrastructure. Traditional monolithic architectures bind hardware resources such as CPUs, accelerators, and memory to a single machine, limiting the flexibility of resource management. In traditional cloud data centers, the deployment unit is a monolithic server, which includes all the hardware resources necessary to run applications (primarily CPU, RAM, network, and disk). Traditional monolithic architectures have two main limitations: difficulty in achieving full resource utilization and coarse-grained hardware maintenance. Decoupled architectures have emerged as a solution that overcomes these limitations, enabling full resource utilization and coarse-grained hardware maintenance.
[0003] Decoupled architecture addresses these limitations by allowing independent management (maintenance, allocation) of each resource. Memory and hardware accelerators can be dynamically allocated to processing units to improve application performance, while high-speed, low-latency electrical and optical interconnects are prerequisites for realizing the data center decoupling concept. The goal of decoupled architecture is to decouple compute, storage / memory, and network elements within a single machine. Typically, due to its distributed and heterogeneous nature, decoupled architecture is not well-suited for traditional Trusted Execution Environment (TEE) technologies. Some well-known cloud service providers offer confidential cloud computing platforms that use TEEs to protect sensitive data. However, decoupled architecture presents significant challenges that limit the efficient use of current TEE technologies. Traditional cloud computing systems lack scalability, unified authentication methods, and transparent support for confidential virtual machines on decoupled architectures.
[0004] Traditionally, cloud computing platform services include remote authentication mechanisms, enabling customers to remotely verify the security and integrity of firmware, software, and hardware configurations deployed on the cloud for a target service. Modern cloud workloads, such as AI applications—e.g., services based on large language models—typically handle large amounts of confidential and security-sensitive private data, including medical data, voice / video recordings, and even financial information. To protect the vast amounts of data at work, it is crucial to ensure that machine learning models, computations, and queries are not exposed to or manipulated by any unauthorized parties, including software running by other tenants, OS / hypervisors, devices, and network infrastructure. Therefore, data transferred between distributed heterogeneous computing and storage components is protected and authenticated for its authenticity and integrity before any component participates in any computation. For decades, TEEs have been researched to ensure code integrity and provide strong isolation from untrusted parties. TEEs provide hardware-assisted isolated environments, known as "enclaves," for executing security-sensitive workloads where application code and data are protected from other cloud tenants and applications, and even from compromised privileged software.
[0005] Furthermore, cloud computing platforms do not provide confidential computing services on a discrete architecture. While TEEs provide necessary security features in traditional cloud computing systems, they are not well-suited for discrete architectures because they are device-specific and their protection domain is limited to the target device. Discrete architectures deploy user code and data across various hardware components, making it difficult for device-specific TEEs to protect all distributed components of an application. Therefore, in cloud computing platforms, there is a technical challenge of how to protect distributed components across discrete resource architectures.
[0006] Therefore, based on the above discussion, it is necessary to overcome the aforementioned drawbacks associated with traditional methods and systems for creating confidential cloud computing platforms in a split resource architecture, which includes at least one confidential VM among multiple confidential virtual machines (VMs). Summary of the Invention
[0007] This invention provides a method and system for creating a confidential cloud computing platform in a split-resource architecture, the confidential cloud computing platform comprising at least one confidential VM among a plurality of confidential virtual machines (VMs). This invention provides a solution to address the existing problem of how to protect distributed components across a split-resource architecture in a cloud computing platform. The object of this invention is to provide a solution that at least partially overcomes the problems encountered in the prior art and provides an improved method and system for creating a confidential cloud computing platform in a split-resource architecture (e.g., a confidential split-resource architecture), the confidential cloud computing platform comprising at least one confidential VM among a plurality of confidential virtual machines (VMs).
[0008] One or more objects of the invention are achieved by means of the solutions provided in the appended independent claims. Advantageous embodiments of the invention are further defined in the dependent claims.
[0009] In one aspect, the present invention provides a method for creating a confidential cloud computing platform in a split-resource architecture, the confidential cloud computing platform including at least one confidential VM from a plurality of confidential virtual machines (VMs). The method includes: receiving a security policy from a user at an authentication and key management system, wherein the security policy includes a public key to be used for at least one confidential VM of the user. The method further includes: receiving a request from the user at a split-resource cloud gateway to create at least one confidential VM based on the received security policy; and receiving the request from the cloud gateway at a cloud scheduler system to create at least one confidential VM on the split-resource architecture. Furthermore, the method includes: the cloud scheduler system interacting with monitoring agents in a plurality of compute pools including a plurality of processor hardware resources and a plurality of storage pools including a plurality of storage hardware resources to obtain availability information about the processor hardware resources and the storage hardware resources; and the cloud scheduler system allocating the processor hardware resources and storage hardware resources to the at least one confidential VM based on the results, thereby creating the at least one confidential VM. Additionally, the authentication and key management system performs authentication on the allocated processor hardware resources and the allocated storage hardware resources.
[0010] Advantageously, the method is used to build confidential computing services (i.e., confidential VMs) with a split architecture, enabling users to run existing applications while maintaining the confidentiality of Internet Protocol (code, data) information. Furthermore, the method provides an end-to-end workflow for creating and authenticating confidential VMs on a split architecture. Security policies are used to maintain the data security and confidentiality of at least one confidential virtual machine (VM). Additionally, an authentication and key management system serves as a centralized system for managing security policies, ensuring data integrity and security by verifying security standards through security policies and providing guarantees against any unauthorized access. The interaction between the authentication and key management system, the cloud scheduler system, and the monitoring agent ensures seamless integration of security measures with resource allocation decisions, which facilitate a cohesive workflow within the confidential cloud computing platform for creating at least one confidential VM among multiple confidential virtual machines (VMs). A split resource cloud gateway provides a user-friendly interface for initiating the creation of at least one confidential VM among multiple confidential VMs based on security policies. The cloud scheduler system is used to allocate resources through dynamic interaction with the monitoring agent to optimize resource utilization in real-time or near real-time. Therefore, the method is used to provide efficient and reliable resource utilization, as well as centralized security management, robust authentication and key management systems, to provide a confidential cloud environment, enhance data security, and improve user experience.
[0011] In one implementation, the user application runs in at least one confidential virtual machine (VM) in the confidential cloud computing platform according to a split resource architecture.
[0012] The method improves the confidentiality and integrity of user application code and data by running user applications in at least one confidential virtual machine (VM) in a confidential cloud computing platform. This is achieved by encapsulating the user application within the confidential VM and leveraging the protection features of a TEE to ensure security.
[0013] In another implementation, the confidential VM comprises a separate operating system.
[0014] Advantageously, including a separate operating system within a confidential virtual machine (VM) allows for independent management of resources within the confidential VM. Independent management of resources within the confidential VM promotes efficient resource allocation, optimal utilization, and adaptability to different workloads.
[0015] In another implementation, at least one of the plurality of storage hardware resources and at least one of the plurality of processor hardware resources includes a trusted execution environment (TEE) module.
[0016] Including a TEE element in at least one of the multiple storage hardware resources and at least one of the multiple processor hardware resources enhances the overall security of the confidential cloud computing platform by providing a dedicated secure enclave for critical operations. The TEE element ensures the confidentiality and integrity of data and computation, protects at least one of the multiple storage hardware resources and at least one of the multiple processor hardware resources from unauthorized access or tampering, and safeguards the entire platform from potential threats.
[0017] In another implementation, the TEE element performs measurements on the software components of at least one of the plurality of storage hardware resources including the TEE element and at least one of the plurality of processor hardware resources, and uses the measurements to create a certificate for its corresponding hardware resource, and sends the certificate to the authentication and key management system.
[0018] This system performs measurements on software components of at least one storage hardware resource and at least one processor hardware resource within a confidential cloud computing platform. It also creates credentials for these software components and sends these credentials to an authentication and key management system. This provides enhanced security and integrity guarantees for at least one storage hardware resource and at least one processor hardware resource within a confidential cloud computing platform. When using measurement credentials for software components, the TEE element ensures the verifiable trustworthiness of the relevant software components to prevent unauthorized access or tampering and provides a secure architecture.
[0019] In another implementation, a security controller is provided as part of a TEE element, and the security controller performs measurements on the software components of at least one of the plurality of storage hardware resources and at least one of the plurality of processor hardware resources, excluding the TEE element, and uses the measurements to create a certificate for the corresponding hardware resource, and sends the certificate to the authentication and key management system.
[0020] The integration of the security controller extends security measures to at least one storage hardware resource out of multiple storage hardware resources and at least one processor hardware resource out of multiple processor hardware resources, excluding TEE elements. After measuring the software components, the security controller generates a certificate specific to each hardware resource. This certificate is then transmitted to the authentication and key management system, which contributes to the overall security of the storage and processor hardware resources that are not equipped with TEEs.
[0021] In another implementation, the authentication and key management system performs authentication on at least one of the plurality of storage hardware resources and at least one of the plurality of processor hardware resources.
[0022] By performing authentication on at least one of the storage hardware resources and at least one of the processor hardware resources within a confidential cloud computing platform, the authentication and key management system provides robust security guarantees for at least one of the storage hardware resources and at least one of the processor hardware resources within a confidential cloud computing platform.
[0023] In another implementation, the authentication and key management system provides certificates to at least one of the plurality of storage hardware resources and at least one of the plurality of processor resources based on the execution of the authentication.
[0024] The authentication and key management system issues certificates to at least one storage hardware resource and at least one processor resource based on the execution of authentication, establishing verifiable and documented proof of the integrity and security status of the storage hardware resource and the processor hardware resource.
[0025] In another implementation, at least one of the plurality of storage hardware resources and at least one of the plurality of processor hardware resources communicate with each other using the certificate provided by the authentication and key management system.
[0026] By using certificates provided by the authentication and key management system as an authentication means, the authentication and key management system ensures that communication between storage hardware resources and processor hardware resources is not only encrypted but also verified to prevent unauthorized access and tampering.
[0027] In another aspect, a system is provided, comprising means for performing all steps of the method according to any of the preceding method claims.
[0028] The disclosed system achieves all the advantages and technical effects of the method for creating a confidential cloud computing platform of the present invention.
[0029] It should be understood that all of the above implementation methods can be combined.
[0030] It should be noted that all devices, elements, circuits, units, and components described in this application can be implemented by software or hardware elements or any combination thereof. All steps performed by the various entities described in this application, and the functions to be performed by the various entities described, are intended to refer to the respective entities performing the respective steps and functions. Although the specific functions or steps performed by external entities are not reflected in the detailed descriptions of the specific elements of the entities performing the specific steps or functions in the following description of specific embodiments, those skilled in the art will understand that these methods and functions can be implemented by corresponding software or hardware elements or any combination thereof. It is understood that the features of the present invention are readily combined in various ways without departing from the scope of the invention as defined by the appended claims.
[0031] Additional aspects, advantages, features, and objects of the invention will become apparent from the accompanying drawings and the detailed description of illustrative implementations as explained in conjunction with the following appended claims. Attached Figure Description
[0032] The above-described invention and the following detailed description of illustrative embodiments can be better understood when read in conjunction with the accompanying drawings. Exemplary structures of the invention are shown in the drawings to illustrate the invention. However, the invention is not limited to the specific methods and tools disclosed herein. Furthermore, those skilled in the art will understand that the drawings are not drawn to scale. Where possible, the same elements are represented by the same numbers.
[0033] The following figures will now be used as examples to describe embodiments of the present invention, wherein: Figure 1 This is a diagram illustrating at least one confidential virtual machine among multiple confidential virtual machines in a split architecture, provided by an embodiment of the present invention; Figure 2 This is a diagram illustrating the interaction between untrusted execution environment elements and trusted execution environment elements, provided by an embodiment of the present invention. Figure 3 This is an illustration of a system for creating a confidential cloud computing platform, as described in an embodiment of the present invention, the confidential cloud computing platform including at least one confidential VM among a plurality of confidential virtual machines (VMs); Figure 4 This is a diagram illustrating remote authentication of the hardware and storage resources of at least one confidential VM among a plurality of confidential virtual machines (VMs) provided in an embodiment of the present invention. Figure 5This is a flowchart illustrating a method for creating a confidential cloud computing platform in a split resource architecture, as provided in an embodiment of the present invention. The confidential cloud computing platform includes at least one confidential VM among a plurality of confidential virtual machines (VMs).
[0034] In the accompanying diagrams, underlined numbers indicate the item in which the underlined number appears or the item adjacent to the underlined number. Ununderlined numbers relate to the item identified by the line that associates the ununderlined number with the item. When a number is ununderlined and has an associated arrow, the ununderlined number is used to identify the general item that the arrow points to. Detailed Implementation
[0035] The following detailed description illustrates embodiments of the present invention and ways in which these embodiments can be implemented. While some modes of implementing the invention have been disclosed, those skilled in the art will recognize that other embodiments for implementing or practicing the invention may also exist.
[0036] Figure 1 This is a diagram illustrating at least one confidential virtual machine among multiple confidential virtual machines in a split architecture, as provided in an embodiment of the present invention. (Reference) Figure 1 The diagram illustrates a block diagram 100 including a compute pool 102 and a storage pool 104, which are interconnected via a high-speed network 108 and at least one of a plurality of confidential virtual machines (VMs) 110, said confidential VM being connected to another virtual machine (VM) 112. At least one of the confidential VMs 110 includes a hypervisor 136.
[0037] Computation pool 102 refers to a collection of multiple processor hardware resources organized in a separate manner, serving as a dynamic and flexible resource allocation system within a confidential cloud computing platform. Furthermore, computation pool 102 includes a central processing unit (CPU) pool 114, an XPU pool 116, and a field-programmable gate array (FPGA) pool 118. CPU pool 114 refers to processor hardware resources designed to meet general-purpose computing needs. XPU pool 116 refers to accelerators (e.g., general-purpose processing units (GPUs), neural processing units (NPUs), etc.) within computation pool 102 that go beyond traditional central processing units (CPUs) to meet the specific computing needs of users within the confidential cloud computing platform. Additionally, FPGA pool 118 refers to processor hardware resources used to execute specialized algorithms or tasks.
[0038] Storage pool 104 is a collection of multiple storage hardware resources used as a centralized dynamic resource management system in a confidential cloud computing platform to meet users' data storage needs. Storage pool 104 includes a memory (MEM) pool 120, a solid-state drive (SSD) pool 122, and a hard disk drive (HDD) pool 124.
[0039] High-speed network 108 refers to a communication infrastructure capable of rapidly and efficiently transmitting data between different hardware resources and storage resources (such as the processor hardware resources of computing pool 102 and the storage hardware resources of storage pool 104). High-speed network 108 is capable of transmitting data at rates exceeding typical network speeds, providing the necessary bandwidth to meet the needs of modern data-intensive applications.
[0040] At least one confidential virtual machine among the multiple confidential virtual machines 110 and another VM 112 refer to virtual machines used to provide a secure and confidential execution environment for user application 126 and data to user 106. Furthermore, at least one confidential VM among the multiple confidential VMs 110 includes a virtual central processing unit (vCPU) 128, a virtual general purpose unit (vGPU) 130, a vFPGA 132, and a vMEM 134. At least one confidential VM among the multiple confidential VMs 110 also includes a user application 126, which is software or a program executing within at least one confidential VM among the multiple confidential VMs 110 in the confidential cloud computing platform, and includes tasks and computations initiated by user 106 in a secure and isolated environment.
[0041] In one implementation scenario, to build confidential computing services in a split architecture, existing trusted execution environments (TEEs) cannot be applied to monolithic servers due to the distributed nature and device heterogeneity of the split resource architecture. Furthermore, existing TEEs are device-specific, and their protection domains are limited to certain target devices. Additionally, the split resource architecture deploys user code and data across an arbitrary set of multiple hardware resources (i.e., compute pool 102 and storage pool 104), making it difficult to protect all distributed hardware resources of the user application. The hypervisor 136 acts as an intermediary layer between physical hardware (e.g., first hardware 138A, second hardware 138B, third hardware 138C, and fourth hardware 138D) and virtualization components. For example, the hypervisor 136 acts as an intermediary layer between vCPU 128 and CPU 140, vGPU 130 and XPU 142, vFPGA 132 and FPGA 144, and vMEM 134 and MEM 146. Typically, due to the heterogeneity of a decoupled architecture, it is difficult to ensure end-to-end security for user applications 126 deployed across different hardware resources. Furthermore, multiple processor hardware resources (e.g., CPU pool 114, XPU pool 116, and FPGA pool 118) and multiple storage hardware resources (e.g., MEM pool 120, SSD pool 122, and HDD pool 124) have different roots of trust and include both TEE and non-TEE devices, which presents challenges in remote authentication and establishes a trusted environment for building at least one confidential VM among the multiple confidential VMs 110 in a decoupled architecture. Therefore, the decoupled resource architecture provides high flexibility in resource allocation, supporting the allocation and reallocation of resources as needed, improving resource utilization by decoupling resources from specific servers, and thus achieving more efficient resource utilization. In addition, the decoupled resource architecture facilitates the rapid adoption of next-generation components in the data center through functional separation, achieving 10 to 100 times larger CPU and memory capacity and 100 times greater bandwidth than server-centric architectures, thereby enhancing the overall capacity and overall performance of at least one confidential VM among the multiple confidential VMs 110.
[0042] Figure 2 This is a diagram illustrating the interaction between untrusted execution environment elements and trusted execution environment elements, provided by an embodiment of the present invention. (See reference) Figure 2An exemplary diagram 200 is shown, describing a trusted execution environment (TEE) module 202, which includes a security controller 204 and an authentication and access controller 206. The TEE element 202 interacts with a non-TEE FPGA 218, a non-TEE GPU 220, and a non-TEE MEM 222.
[0043] In one implementation scenario, TEE element 202 (e.g., AMD SEV SNP, ARM CCA, etc.) refers to a secure area that helps protect the code and data loaded within the processor from unauthorized access or modification. Furthermore, TEE element 202 is used to maintain the confidentiality, integrity, and isolation of trusted user application code and data on the device, resisting tampering and disclosure from software or physical attacks, and even from privileged code. Security controller 204 is used to enable security features for non-TEE elements (i.e., non-TEE FPGA 218, non-TEE GPU 220, and non-TEE MEM 222). Security controller 204 acts as an intermediary between TEE element 202 and non-TEE elements to provide secure boot, isolation, trusted paths, and remote authentication to non-TEE elements. Security controller 204 is a controller with a hardware root of trust that can isolate (i.e., through forced isolation 212) its own memory to ensure that each non-TEE element is assigned to a single tenant and is not shared with any other tenants or services in the discrete architecture. A security controller 204 configured within TEE element 202 is used to protect communication between TEE element 202 and non-TEE elements by performing authenticated encryption and decryption 214 using a shared secret setting during user setup, and to enforce access control by checking whether the source and destination belong to the same user application. Furthermore, the security controller 204 accordingly supports or blocks communication between the source and destination (i.e., between TEE element 202 and non-TEE elements). The security controller 204 is physically connected to the non-TEE elements and creates a locally unique identifier for each non-TEE element during startup. Additionally, the security controller 204 maintains input and output buffers for the non-TEE elements and allocates non-overlapping areas to ensure isolated access. Furthermore, the security controller 204 tracks non-TEE elements belonging to the same application or confidential virtual machine to enforce access control. In one implementation, the security controller 204 runs within TEE element 202 and can be accessed by a user (e.g., Figure 1User authentication (106). In one implementation, authentication and access controller 206 manages and controls the authentication and access permissions of entities interacting with TEE element 202, for example, by verifying the identity of the user or component seeking access to TEE element 202 to ensure that only authorized entities are granted access. Furthermore, scheduling 216 helps to adjust and manage the execution of tasks and jobs outside of TEE elements. Typically, to securely run applications in the cloud, users run applications within a secure enclave using confidential computing services. Therefore, a remote authentication mechanism is needed to support user authentication for the execution of applications deployed in remote environments. Traditionally, remote authentication mechanisms provide TEE element 202 with the integrity of the application at the start of the enclave. However, applications dynamically load libraries and configurations during runtime, which can add vulnerabilities to the application that TEE element 202 cannot further detect. Therefore, it is necessary to measure the dynamic behavior of the application. Furthermore, TEE element 202 provides different formats for authentication reports, which requires converting all different formats to a unified authentication format. Measurement functionality involves systematically evaluating and recording various aspects that define the behavior and integrity of applications or elements within a TEE element. Measurement 208 and Authentication 210 capture a comprehensive snapshot of the runtime environment, including libraries and configurations dynamically loaded during application execution, to ensure that measurements accurately reflect the current and dynamic state of the system. Furthermore, the authentication function utilizes measurement data to establish the integrity and legitimacy of TEE element 202 to external entities and supports users or external systems in verifying the confidentiality and integrity of applications running within TEE element 202 to prevent potential attacks or unauthorized interference. Authentication 210 acts as a means for users to trust and verify the secure execution of applications in remote environments, providing assurance that TEE element 202 operates as intended and is not compromised.
[0044] Figure 3 The diagram illustrates a system for creating a confidential cloud computing platform, as described in an embodiment of the present invention. This confidential cloud computing platform includes at least one confidential VM from a plurality of confidential virtual machines (VMs). Combined with... Figure 1 and Figure 2 Component description Figure 3 . refer to Figure 3The diagram shows a system 300, which includes a user 302, an authentication and key management system 306, a split resource cloud gateway 310, a cloud scheduler system 314, a split operating system 320, and a security system 322 including a trusted execution environment (TEE), a hardware security module (HSM), and a trusted platform module (TPM).
[0045] The authentication and key management system 306 is used to process and manage decryption and encryption keys for functions used in the enclave. Furthermore, the authentication and key management system 306 is also used to store keys and, after verifying the keys, to provide them to at least one confidential VM among a plurality of confidential virtual machines (VMs), such as at least one confidential VM 110 running user application 126.
[0046] A system 300 is provided for creating a confidential cloud computing platform in a split resource architecture, the confidential cloud computing platform including at least one confidential virtual machine among a plurality of confidential virtual machines 110.
[0047] The authentication and key management system 306 receives a security policy from user 302. The security policy 304 includes a public key for at least one confidential VM 110 used by the user. The security policy 304 serves as a set of guidelines and configurations outlining how the security features of at least one confidential VM 110 should be managed. The public key is used to perform encryption and decryption within the confidential VM to ensure the confidentiality and integrity of user application code and data of user application 126. The authentication and key management system 306 utilizes the security policy 304 to enforce security measures and manage encryption keys to provide a secure environment for the operation of at least one confidential VM 110.
[0048] Furthermore, the split resource cloud gateway 310 is used to receive requests from user 302 to create at least one confidential VM 110 according to the received security policy 304, for example at operation 308, or at operation 312. The split resource cloud gateway 310 acts as an intermediary, responsible for routing requests and managing the interaction between user 302 and the cloud computing platform infrastructure. The split resource cloud gateway 310 efficiently routes requests according to security policy 304 to create at least one confidential VM among multiple confidential VMs 110. Therefore, the split resource cloud gateway 310 provides efficient routing, security policy enforcement, and specialized processing of user requests for creating confidential VMs within a split architecture.
[0049] Furthermore, the cloud scheduler system 314 receives requests from the split-resource cloud gateway 310 to create at least one confidential VM 110 on top of the split-resource architecture. In other words, the cloud scheduler system 314 initiates a request to create at least one confidential VM among multiple confidential VMs 110 on the split-resource architecture. By receiving requests from the split-resource cloud gateway 310, the cloud scheduler system 314 helps orchestrate the deployment of confidential virtual machines (VMs), thereby contributing to the overall functionality and responsiveness of the cloud computing platform.
[0050] Furthermore, the cloud scheduler system 314 interacts with monitoring agents in multiple compute pools 102, which include multiple processor hardware resources, and multiple storage pools 104, which include multiple storage hardware resources, to obtain availability information about the processor and storage hardware resources. The cloud scheduler system 314 is used to collect information about the monitoring agents in the multiple compute pools 102, which include multiple processor hardware resources (e.g., CPU, XPU, GPU, NPU, TPU, or FPGA), and the multiple storage pools 104, which include multiple storage hardware resources (e.g., DRAM, SSD, HDD). In one implementation, at operation 316, the cloud scheduler system 314 is used to launch a new confidential virtual machine, and at operation 318, the cloud scheduler system 310 is used to receive resource updates from the multiple hardware resources and the multiple storage resources. By obtaining real-time availability information, the cloud scheduler system 314 can make informed decisions about where to allocate or reallocate multiple hardware resources based on the current state of the multiple compute pools 102 and the multiple storage pools 104, to optimize resource utilization and enhance the overall performance of the cloud computing platform.
[0051] According to one embodiment, the multiple storage hardware resources include DRAM, SSD, or HDD hardware resources. In one implementation, the multiple storage hardware resources include dynamic random-access memory (DRAM). In another implementation, the multiple storage hardware resources include solid-state drives (SSDs). In yet another implementation, the multiple storage hardware resources include hard disk drives (HDDs). Therefore, system 300 can utilize a variety of different storage hardware resources to achieve different functions; for example, DRAM can be used for volatile and high-speed storage, SSDs can be used for faster and non-volatile storage using flash memory, and HDDs can be used for traditional, larger-capacity, and slower mechanical disk-based storage. Thus, multiple storage hardware resources provide flexibility to meet different performance and capacity requirements within system 300.
[0052] According to one embodiment, the plurality of processor hardware resources include CPU, XPU, GPU, NPU, TPU, or FPGA hardware resources. In one implementation, the plurality of processor hardware resources include a central processing unit (CPU) for processing general computing tasks. In another implementation, the plurality of processor hardware resources include an accelerated processing unit (XPU) for accelerating specific workloads. In yet another implementation, the plurality of processor hardware resources include a general processing unit (GPU) adept at graphics-related computations. In yet another implementation, the plurality of processor hardware resources include a neural processing unit (NPU) for performing optimized neural network tasks. In yet another implementation, the plurality of processor hardware resources include a tensor processing unit (TPU) for machine learning-related computations. In yet another implementation, the plurality of processor hardware resources include a field-programmable gate array (FPGA) for the flexibility of implementing custom hardware acceleration. Similarly, other hardware resources may also be included among the plurality of processor hardware resources without affecting the scope of the invention. Therefore, this inclusion of multiple processor hardware resources provides versatility to meet a variety of computing needs in a broader context of the system 300.
[0053] Furthermore, the cloud scheduler system 314 allocates processor hardware resources and storage hardware resources to at least one confidential VM 110 based on the availability of information regarding processor hardware resources and storage hardware resources, thereby creating at least one confidential VM 110. The cloud scheduler system 314 first receives a request and then collects basic monitoring information regarding the current availability of processor hardware resources and storage hardware resources in compute pool 102 and storage pool 104. After acquiring the information, the cloud scheduler system 314 allocates processor hardware resources and storage hardware resources to at least one confidential VM 110 based on the availability of information regarding processor hardware resources and storage hardware resources, thereby creating at least one confidential VM among multiple confidential VMs 110. Therefore, the cloud scheduler system 314 ensures the reasonable and efficient allocation of hardware resources and supports at least one confidential VM among multiple confidential VMs 110 to adapt to workload changes, thereby enhancing the reliability and responsiveness in meeting user needs.
[0054] According to one embodiment, the confidential VM 110 includes a discrete operating system 320. The discrete operating system 320 (or a lightweight discrete operating system) serves as the underlying software layer responsible for managing the isolation and encryption of user applications 126. The discrete operating system 320 collaborates with a security controller 204 and hardware trusted execution environment (TEE) elements (e.g., first TEE element 334A, second TEE element 334B, third TEE element 334C, and fourth TEE element 334D) connected to multiple compute pools 102 and multiple storage pools 104 to ensure the secure execution of applications within the VM, thereby providing a seamless and user-friendly experience.
[0055] Furthermore, the authentication and key management system 306 performs authentication on the allocated processor hardware resources and allocated storage hardware resources. In other words, the authentication and key management system 306 performs a process of verifying and confirming the integrity and trustworthiness of the allocated processor hardware resources and allocated storage hardware resources within the confidential cloud computing platform, for example, at operation 324. The authentication and key management system 306 initiates an authentication process for each of these resource types and performs authentication by examining various security parameters and measurements associated with the processor hardware elements and storage hardware elements. This authentication may include verifying the integrity of the startup process, ensuring no unauthorized modifications have been made, and confirming that the hardware resources are operating as intended and unmodified by the code.
[0056] According to one embodiment, the authentication and key management system 306 performs authentication on at least one storage hardware resource among a plurality of storage hardware resources and at least one processor hardware resource among a plurality of processor hardware resources. In one implementation, the authentication and key management system 306 performs authentication on at least one storage hardware resource among a plurality of storage hardware resources. In another implementation, the authentication and key management system 306 performs authentication on at least one processor hardware resource among a plurality of processor hardware resources. In yet another implementation, the authentication and key management system 306 performs authentication on at least one storage hardware resource among a plurality of storage hardware resources and at least one processor hardware resource among a plurality of processor hardware resources. Advantageously, the authentication process performed by the authentication and key management system 306 is used to maintain the integrity of at least one confidential virtual machine among a plurality of confidential virtual machines 110 created, and to prevent potential security risks or unauthorized access to confidential VMs and related data.
[0057] According to one embodiment, the authentication and key management system 306 provides a certificate to at least one storage hardware resource and at least one processor resource among a plurality of storage hardware resources in accordance with the execution of authentication. In one implementation, the authentication and key management system 306 provides a certificate to at least one storage hardware resource among a plurality of storage hardware resources in accordance with the execution of authentication on at least one storage hardware resource among a plurality of storage hardware resources. In another implementation, the authentication and key management system 306 provides a certificate to at least one processor hardware resource among a plurality of processor hardware resources in accordance with the execution of authentication on at least one storage hardware resource among a plurality of storage hardware resources and at least one processor hardware resource among a plurality of processor hardware resources. Advantageously, the authentication and key management system 306 providing the certificate helps to establish a verifiable and documented proof of the integrity and security state of at least one storage hardware resource among a plurality of storage hardware resources and at least one processor resource among a plurality of processor resources.
[0058] According to one embodiment, at least one storage hardware resource among a plurality of storage hardware resources and at least one processor hardware resource among a plurality of processor hardware resources communicate with each other using certificates provided by an authentication and key management system 306. Therefore, by using certificates to facilitate secure and trusted communication, at least one storage hardware resource and at least one processor hardware resource participate in communication between the at least one storage hardware resource and the at least one processor hardware resource through the authentication and key management system 306, thereby ensuring the integrity and authenticity of the communicating entities.
[0059] According to one embodiment, at least one storage hardware resource among multiple storage hardware resources and at least one processor hardware resource among multiple processor hardware resources include a Trusted Execution Environment (TEE) element. In one implementation, at least one storage hardware resource among multiple storage hardware resources includes a TEE element. In another implementation, at least one processor hardware resource among multiple processor hardware resources includes a TEE element. In yet another implementation, at least one storage hardware resource among multiple storage hardware resources and at least one processor hardware resource among multiple processor hardware resources include a TEE element. Including a TEE element in at least one storage hardware resource among multiple storage hardware resources and at least one processor hardware resource among multiple processor hardware resources enhances the overall security of the confidential cloud computing platform by providing a dedicated secure enclave for critical operations. The TEE element ensures the confidentiality and integrity of data and computation, protects at least one storage hardware resource among multiple storage hardware resources and at least one processor hardware resource among multiple processor hardware resources from unauthorized access or tampering, and protects the entire platform from potential threats.
[0060] According to one embodiment, a TEE element performs measurements on the software components of at least one storage hardware resource and at least one processor hardware resource among a plurality of storage hardware resources including the TEE element, uses the measurements to create a certificate for its corresponding hardware resource, and sends the certificate to an authentication and key management system 306. In one implementation, the TEE element performs measurements on the software components of at least one storage hardware resource among a plurality of storage hardware resources including the TEE element. In another implementation, the TEE element performs measurements on the software components of at least one processor hardware resource among a plurality of processor hardware resources including the TEE element. In yet another implementation, the TEE element performs measurements on the software components of at least one storage hardware resource and at least one processor hardware resource among a plurality of storage hardware resources including the TEE element. Performing measurements on the software components of at least one storage hardware resource and at least one processor hardware resource among a plurality of processor hardware resources, creating a certificate for its corresponding hardware resource, and sending the certificate to the authentication and key management system 306 provides enhanced security and integrity guarantees for at least one storage hardware resource and at least one processor hardware resource among a plurality of storage hardware resources within a confidential cloud computing platform. For example, automated and transparent remote authentication can be achieved by sending a notification to the authentication and key management system 306, such as at operation 328. When using measurement proofs of software components, the TEE element ensures the verifiable trustworthiness of the relevant software components to prevent unauthorized access or tampering and provides a secure architecture.
[0061] According to one embodiment, at least one storage hardware resource among multiple storage hardware resources and at least one processor hardware resource among multiple processor hardware resources do not include a TEE element. In one implementation, at least one storage hardware resource among multiple storage hardware resources does not include a TEE element. In another implementation, at least one processor hardware resource among multiple processor hardware resources does not include a TEE element. In yet another implementation, at least one storage hardware resource among multiple storage hardware resources and at least one processor hardware resource among multiple processor hardware resources do not include a TEE element. Because of the absence of a TEE element, the cloud computing platform introduces versatility to support hardware resources that do not have a TEE element, while still being part of the cloud computing platform. Flexibility is advantageous when handling different applications and workloads that may not require the security measures provided by a TEE element. Therefore, the cloud computing platform can cater to diverse computing needs, optimize resource allocation based on the specific security requirements of different tasks, thereby enhancing scalability and adaptability, making the confidential cloud computing platform suitable for a wider range of applications, while maintaining a robust security framework when needed, and helping to achieve a balanced and flexible approach within the confidential cloud computing platform.
[0062] According to one embodiment, a security controller is provided as part of a TEE element, and the security controller performs measurements on software components of at least one storage hardware resource and at least one processor hardware resource among a plurality of storage hardware resources excluding the TEE element, and uses the measurements to create a certificate for the corresponding hardware resource, and sends the certificate to an authentication and key management system 306. In one implementation, the TEE element performs measurements on software components of at least one storage hardware resource among a plurality of storage hardware resources excluding the TEE element. In another implementation, the TEE element performs measurements on software components of at least one processor hardware resource among a plurality of processor hardware resources excluding the TEE element. In yet another implementation, the TEE element performs measurements on software components of at least one storage hardware resource and at least one processor hardware resource among a plurality of storage hardware resources excluding the TEE element. The integration of the security controller 204 extends security measures to at least one storage hardware resource among a plurality of storage hardware resources and at least one processor hardware resource among a plurality of processor hardware resources excluding the TEE element. After measuring the software components, the security controller 204 generates a certificate specific to each hardware resource. Then, for example at operation 326, the proof is transmitted to the authentication and key management system 306, which contributes to the overall security of storage hardware resources and processor hardware resources that are not equipped with TEE.
[0063] According to one embodiment, a user application runs in at least one confidential VM 110 within a confidential cloud computing platform, based on a split-resource architecture. In one implementation, the user application runs via a secure shell (SSH) within at least one confidential VM 110 in the confidential cloud computing platform, for example, at operation 330. By running the user application within at least one confidential virtual machine (VM) in the confidential cloud computing platform, the cloud computing platform enhances the confidentiality and integrity of the user application's code and data by encapsulating the user application within at least one confidential VM among multiple confidential VMs 110, leveraging the protection features of a TEE to ensure security.
[0064] Figure 4 This is a diagram illustrating remote authentication of the hardware and storage resources of at least one confidential VM among a plurality of confidential virtual machines (VMs), provided by another embodiment of the present invention. (In conjunction with...) Figure 1 , Figure 2 and Figure 3 Component description Figure 4 . refer to Figure 4The diagram shows Figure 400, which includes a computing pool 102 and a storage pool 104 connected via a high-speed network 108.
[0065] In one implementation scenario, the authentication and key management system 306 handles all the complexities of authentication for different heterogeneous TEEs within a split architecture on behalf of the user (i.e., user 302). The authentication and key management system 306 provides a common and unified authentication abstraction for multiple heterogeneous hardware resources in the split architecture by providing certificates / tokens for each element belonging to at least one of the multiple confidential VMs 110, and the TEE elements communicate directly with each other using certificates (or tokens) over a secure network (e.g., via transport layer security (TLS)). Furthermore, multiple compute pools 102, multiple storage pools 104, and the authentication and key management system 306 are connected to a high-speed network 108 via a security system 322 comprising a trusted execution environment (TEE), a hardware security module (HSM), and a trusted platform module (TPM).
[0066] The authentication and key management system 306 is used to check authentication offers and issue tokens or certificates upon successful verification, along with necessary configuration and secret settings, such as encryption and decryption keys. Furthermore, the authentication and secret settings 402 significantly enhance security and trust within the decoupled architecture. Additionally, the authentication and secret settings 402 ensure that only trusted and authenticated hardware resources and storage resources can participate in secure communication, thereby reducing the risk of unauthorized access and fostering a highly secure and trusted system environment.
[0067] Figure 5 This is a flowchart illustrating a method for creating a confidential cloud computing platform in a split resource architecture, as provided in an embodiment of the present invention. The confidential cloud computing platform includes at least one confidential VM from a plurality of confidential virtual machines (VMs). Reference Figure 5 The diagram illustrates a flowchart of a method 500 for creating a confidential cloud computing platform in a split resource architecture, the confidential cloud computing platform including at least one confidential virtual machine from a plurality of confidential virtual machines 110. Method 500 includes steps 502 and 512.
[0068] A method 500 is provided for creating a confidential cloud computing platform in a split resource architecture, the confidential cloud computing platform including at least one confidential virtual machine among a plurality of confidential virtual machines 110.
[0069] At step 502, method 500 includes receiving a security policy 304 from user 302 at authentication and key management system 306. Furthermore, security policy 304 includes a public key for at least one confidential VM 110 to be used by the user.
[0070] At step 504, method 500 includes: receiving a request from user 302 at a split resource cloud gateway 310 to create at least one confidential VM 110 according to the received security policy 304.
[0071] At step 506, method 500 includes: receiving the request at cloud scheduler system 314 from split resource cloud gateway 310 to create at least one confidential VM 110 on the split resource architecture.
[0072] At step 508, method 500 includes interacting with monitoring agents in multiple computing pools 102, which include multiple processor hardware resources, and multiple storage pools 104, which include multiple storage hardware resources, to obtain availability information about the processor hardware resources and storage hardware resources through cloud scheduler system 314.
[0073] At step 510, method 500 includes: cloud scheduler system 314 allocating processor hardware resources and storage hardware resources to at least one confidential VM 110 based on the availability of information about processor hardware resources and storage hardware resources, thereby creating at least one confidential VM 110. Furthermore, authentication and key management system 306 performs authentication on the allocated processor hardware resources and allocated storage hardware resources.
[0074] At step 512, method 500 includes: running user application 126 in at least one confidential VM 110 in a confidential cloud computing platform according to a split resource architecture.
[0075] Advantageously, method 500 is used to build confidential computing services (i.e., confidential VMs) with a split architecture, enabling users to run existing applications while maintaining the confidentiality of Internet protocol (code, data). Furthermore, method 500 also provides an end-to-end workflow for creating and authenticating confidential VMs on a split architecture. Security policies are used to maintain the data security and confidentiality of at least one confidential virtual machine (VM). Additionally, authentication and key management system 306 serves as a centralized system for managing security policies to ensure data integrity and security by verifying security standards through security policies and providing guarantees against any unauthorized access. The interaction between the authentication and key management system, cloud scheduler system 314, and monitoring agent ensures seamless integration of security measures with resource allocation decisions, which facilitate a cohesive workflow within the confidential cloud computing platform for creating at least one confidential VM among multiple confidential VMs 110. Split resource cloud gateway 310 provides a user-friendly interface for initiating the creation of at least one confidential VM among multiple confidential VMs based on security policies. The cloud scheduler system 314 is used to allocate resources through dynamic interaction with a monitoring agent to optimize resource utilization in real time or near real time. Therefore, method 500 provides efficient and reliable resource utilization, as well as centralized security management and a robust authentication and key management system 306, to provide a confidential cloud environment, enhance data security, and improve user experience.
[0076] A computer program is provided, including instructions that, when executed by a computer system, cause the computer system to implement the method 500. In one example, the instructions are implemented on a computer-readable medium, including but not limited to electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), read-only memory (ROM), hard disk drive (HDD), flash memory, secure digital card (SD card), solid-state drive (SSD), computer-readable storage media, and / or CPU cache memory. In one example, the instructions are generated by a computer program implemented for method 500, which is used to create a confidential cloud computing platform in a split-resource architecture, the confidential cloud computing platform including at least one confidential virtual machine among a plurality of confidential virtual machines 110.
[0077] Modifications to the embodiments of the invention described above may be made without departing from the scope of the invention as defined by the appended claims. The terms “comprising,” “combining,” “having,” “is,” and the like used to describe and claim the invention are intended to be interpreted in a non-exclusive manner, allowing for the presence of items, components, or elements not explicitly described. Singular references should also be interpreted as relating to the plural. The term “exemplary” as used herein means “as an example, instance, or illustration.” Any embodiment described as “exemplary” is not necessarily to be construed as being more preferred or advantageous than other embodiments, or excluding combinations of features from other embodiments. The term “optionally” as used herein means “provided in some embodiments and not in others.” It should be understood that certain features of the invention described in the context of a single embodiment for clarity may also be provided in combination in a single embodiment. Conversely, various features of the invention described in the context of a single embodiment for brevity may also be provided individually or in any suitable combination or as embodiments of any other described aspect of the invention.
Claims
1. A method (500) for creating a confidential cloud computing platform in a split resource architecture, characterized in that, The confidential cloud computing platform includes at least one confidential VM from a plurality of confidential virtual machines (VMs) (110), and the method includes the following steps: (a) Receive a security policy (304) from a user (302) at an authentication and key management system (306), wherein the security policy (304) includes a public key to be used for at least one confidential VM (110) of the user; (b) Receive a request from the user to create at least one confidential VM (110) at the split resource cloud gateway (310) in accordance with the received security policy (304); (c) Receive the request from the split resource cloud gateway (310) at the cloud scheduler system (314) to create at least one confidential VM (110) on the split resource architecture. (d) The cloud scheduler system (314) interacts with monitoring agents in multiple computing pools (102) including multiple processor hardware resources and multiple storage pools (104) including multiple storage hardware resources to obtain availability information about the processor hardware resources and the storage hardware resources. (e) Based on the result of step (d), the cloud scheduler system (314) allocates processor hardware resources and storage hardware resources to the at least one confidential VM (110) to create the at least one confidential VM (110). The authentication and key management system (306) performs authentication on the allocated processor hardware resources and the allocated storage hardware resources.
2. The method (500) according to claim 1, characterized in that, The user application (126) runs in at least one confidential VM (110) in the confidential cloud computing platform according to the split resource architecture.
3. The method (500) according to claim 2, characterized in that, The confidential VM (110) includes a separate operating system (128).
4. The method (500) according to claim 1, characterized in that, At least one of the plurality of storage hardware resources and at least one of the plurality of processor hardware resources includes a trusted execution environment (TEE) element (202).
5. The method (500) according to claim 4, characterized in that, The TEE element (202) performs measurements on the software components of at least one of the plurality of storage hardware resources including the TEE element (202) and at least one of the plurality of processor hardware resources, and uses the measurements to create a certificate for its corresponding hardware resource, and sends the certificate to the authentication and key management system (306).
6. The method (500) according to claim 1, characterized in that, At least one of the plurality of storage hardware resources and at least one of the plurality of processor hardware resources does not include the TEE element (202).
7. The method (500) according to claim 6, characterized in that, A security controller (204) is provided as part of a TEE element (202), and the security controller (204) performs measurements on the software components of at least one of the plurality of storage hardware resources and at least one of the plurality of processor hardware resources, excluding the TEE element, and uses the measurements to create a certificate for the corresponding hardware resource and sends the certificate to the authentication and key management system (306).
8. The method (500) according to claim 5 or 7, characterized in that, The authentication and key management system (306) performs authentication on at least one of the plurality of storage hardware resources and at least one of the plurality of processor hardware resources.
9. The method (500) according to claim 8, characterized in that, The authentication and key management system (306) provides certificates to at least one of the plurality of storage hardware resources and at least one of the plurality of processor resources in accordance with the execution of the authentication.
10. The method (500) according to claim 9, characterized in that, The at least one of the plurality of storage hardware resources and the at least one of the plurality of processor hardware resources communicate with each other using the certificate provided by the authentication and key management system (306).
11. The method (500) according to claim 1, characterized in that, The multiple storage hardware resources include DRAM, SSD, or HDD hardware resources.
12. The method (500) according to claim 1, characterized in that, The multiple processor hardware resources include CPU, XPU, GPU, NPU, TPU, or FPGA hardware resources.
13. A system (300), characterized in that, Includes means for performing all steps of the method according to any of the above methods (500) claims.
14. A computer program, characterized in that, Includes instructions that, when the computer program is executed on a computer system, perform all the steps of the method described in any of the preceding methods (500) claims.
