Service providing method, tenant configuration management method and related device

Abstract

The embodiment of the invention discloses a service providing method, a tenant configuration management method and a related device. And determining a target computing power pool associated with the tenant according to the configuration information of the computing power pool, determining a target data resource pool needing to be connected with the target computing power pool according to the configuration information of the resource pool, and establishing a connection between the target computing power pool and the target data resource pool, so that the target computing power pool can access the target data resource pool based on the connection. The computing power pool and the resource pool are decoupled, so that the computing power pool can dynamically select and connect the target data resource pool to access the required data resource according to the actual demand when executing the task, instead of only accessing the bound resource pool. Due to the fact that the same resource pool can be connected by different computing power pools, when a plurality of computing power pools access the same data resource, the same data resource does not need to be stored in a plurality of resource pools, the data resource only needs to be stored in the same resource pool which is commonly accessed, repeated storage of the same data resource is avoided, and storage resources are saved.

Description

Technical Field

[0001] This application relates to the field of data access control, specifically to a service provision method, a tenant configuration management method, and related apparatus. Background Technology

[0002] A computing power pool refers to the centralized management of multiple computing resources (such as CPUs, GPUs, and memory) to form a unified computing pool. The computing power pool can dynamically schedule and allocate computing resources according to demand to support efficient computing task execution. A resource pool, on the other hand, refers to a system that centrally stores and manages various data resources (such as models, images, and datasets). The resource pool provides a unified resource management interface, supporting resource uploading, sharing, and access control.

[0003] The industry typically uses a method of bundling computing power pools with resource pools for leasing, or bundling computing power pools and resource pools and deploying them privately for a single account. However, these methods tightly couple resource pools with computing power pools, with physical server computing resources and file resources coexisting. This means that when a computing power pool performs its own business services, it can only access the resource pool it is bound to, and cannot access other resource pools. Secondly, if multiple computing power pools access the same data resource, each resource pool bound to the computing power pool needs to store the same data resource, resulting in wasted storage space and increased storage overhead. Summary of the Invention

[0004] This application provides a service provision method, a tenant configuration management method, and related apparatus. By decoupling the computing power pool from the resource pool, the computing power pool can dynamically select and connect to the target data resource pool to access the required data resources according to actual needs when executing tasks.

[0005] The first aspect of this application provides a service provision method, the method comprising:

[0006] In response to a tenant's access request, the target computing pool associated with the tenant is determined based on the computing pool's configuration information.

[0007] The target data resource pool to be connected to by the target computing power pool is determined based on the configuration information of the resource pool; wherein, the configuration information of the resource pool is used to determine the association between the tenant and the data resource pool, or, the configuration information of the resource pool is used to determine the association between the computing power pool and the data resource pool; the configuration information of the resource pool can be adjusted by the tenant; the same data resource pool can be connected to different computing power pools;

[0008] Establish a connection between the target computing power pool and the target data resource pool, utilize the target computing power pool to access the data resources stored in the target data resource pool based on the connection, and provide computing power services to the tenant.

[0009] A second aspect of this application provides a tenant configuration management method, the method comprising:

[0010] In response to tenants' computing pool configuration commands, create or adjust the configuration information of computing pools;

[0011] In response to tenants' resource pool configuration instructions, create or adjust the configuration information of resource pools;

[0012] The configuration information of the computing power pool is used to determine the target computing power pool associated with the tenant; the configuration information of the resource pool is used to determine the association between the tenant and the data resource pool, or the association between the computing power pool and the data resource pool; the computing power pool and the data resource pool are decoupled so that the same data resource pool can be connected to different computing power pools.

[0013] A third aspect of this application provides a system for generating inference services, the system comprising:

[0014] The system comprises a control module, a computing power pool module, and a data resource pool module. The computing power pool module includes at least one computing power pool, and the data resource pool module includes at least one data resource pool. The computing power pool module and the data resource pool module are decoupled to enable the same data resource pool to be connected to different computing power pools.

[0015] The control module is used to execute the method of the first aspect mentioned above, in order to control the computing power pool to connect to the data resource pool in response to the tenant's access request, access the data resources provided by the data resource pool and provide computing power services to the tenant;

[0016] The control module is also configured to execute the method of the second aspect mentioned above, in response to the tenant's computing pool configuration instruction, to determine the computing pool associated with the tenant; and in response to the tenant's resource pool configuration instruction, to determine the association between the tenant and the data resource pool, or to determine the association between the computing pool and the data resource pool.

[0017] A fourth aspect of this application provides a service providing apparatus, including:

[0018] The computing power pool association module is used to respond to a tenant's access request and determine the target computing power pool associated with the tenant based on the computing power pool's configuration information.

[0019] The data resource pool association module is used to determine the target data resource pool that the target computing power pool needs to connect to based on the resource pool configuration information; wherein, the resource pool configuration information is used to determine the association relationship between the tenant and the data resource pool, or to determine the association relationship between the computing power pool and the data resource pool; the resource pool configuration information can be adjusted; the same data resource pool can be connected to different computing power pools;

[0020] The connection module is used to establish a connection between the target computing power pool and the target data resource pool, utilize the target computing power pool to access the data resources stored in the target data resource pool based on the connection, and provide computing power services to the tenant.

[0021] A fifth aspect of this application provides a tenant configuration management module, including:

[0022] The computing pool configuration module is used to create or adjust the configuration information of computing pools in response to the tenant's computing pool configuration instructions.

[0023] The data resource pool configuration module is used to respond to tenants' resource pool configuration commands and create or adjust the configuration information of resource pools.

[0024] The configuration information of the computing power pool is used to determine the target computing power pool associated with the tenant; the configuration information of the resource pool is used to determine the association between the tenant and the data resource pool, or the association between the computing power pool and the data resource pool; the computing power pool and the data resource pool are decoupled so that the same data resource pool can be connected to different computing power pools.

[0025] A sixth aspect of this application provides a computer device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the methods of the first and second aspects described above.

[0026] A seventh aspect of this application provides a computer storage medium storing instructions that, when executed on a computer, cause the computer to perform the methods described in the first and second aspects.

[0027] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0028] By decoupling the computing power pool from the resource pool, the computing power pool can dynamically select and connect to the target data resource pool to access the required data resources based on actual needs when executing tasks, instead of being limited to accessing only the resource pool it is bound to. Furthermore, since the same resource pool can be connected to by different computing power pools, when multiple computing power pools need to access the same data resource, it is not necessary to store the same data resource in multiple resource pools; the data resource only needs to be stored in the common resource pool, avoiding duplicate storage of the same data resource and saving storage resources.

[0029] Secondly, resource pool permissions are configured on a tenant-by-tenant basis. The same computing power pool can adjust the resource pools it is connected to, realizing flexible scheduling of computing power and resources and avoiding single-point bottlenecks in computing power. That is, when a tenant's high-concurrency tasks are concentrated in a single computing power pool, the resource pools connected to that computing power pool can be dynamically switched to divert some data access requests to resource pool nodes with lower loads. At the same time, other idle computing power pools can be scheduled to connect to the original resource pool, thereby balancing the load pressure of computing power and storage. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the network framework in an embodiment of this application;

[0031] Figure 2 This is a flowchart illustrating the service provision method in the embodiments of this application;

[0032] Figure 3 This is an exemplary schematic diagram illustrating one application scenario of the service provision method in the embodiments of this application;

[0033] Figure 4 This is an exemplary schematic diagram illustrating another application scenario of the service provision method in the embodiments of this application;

[0034] Figure 5 This is an exemplary schematic diagram illustrating another application scenario of the service provision method in the embodiments of this application;

[0035] Figure 6 This is an exemplary schematic diagram illustrating the shared and dedicated scenarios of the computing power pool in the embodiments of this application;

[0036] Figure 7 This is an exemplary schematic diagram illustrating another application scenario of the service provision method in the embodiments of this application;

[0037] Figure 8 This is a flowchart illustrating the tenant configuration management method in this application embodiment;

[0038] Figure 9 This is a schematic diagram of a computer device in an embodiment of this application. Detailed Implementation

[0039] This application provides a service provision method, a tenant configuration management method, and related apparatus. By decoupling the computing power pool from the resource pool, the computing power pool can dynamically select and connect to the target data resource pool to access the required data resources according to actual needs when executing tasks.

[0040] Please see Figure 1 The network framework in this embodiment includes:

[0041] Multiple computing power pools and multiple resource pools can establish communication links between computing power pools and resource pools based on connection protocols to access data.

[0042] A computing power pool comprises multiple computing devices, such as servers, workstations, and high-performance computing nodes, that provide computing resources like CPUs, GPUs, and memory to perform data processing tasks. These devices are connected via a network to form a computing power pool, collectively providing computing services. It's important to note that the multiple computing devices in a pool can be of the same type or different types. A computing power pool can be a single computing cluster or a pool composed of multiple clusters, such as a distributed computing power pool formed by integrating computing clusters from different regions and network environments through virtual network technology.

[0043] A resource pool consists of multiple storage nodes. Each storage node can be deployed with storage devices such as disk arrays and distributed file systems to store various data resources such as model files, datasets, and image packages. Storage nodes synchronize and share resources through a high-speed network, forming a unified resource access point. Similarly, a resource pool can be a single storage cluster or a distributed resource pool composed of multiple storage clusters. For example, storage nodes deployed across regions can be integrated into a unified resource pool using cloud storage technology.

[0044] The connection protocol between the computing power pool and the resource pool supports multiple communication standards such as TCP / IP and RDMA. It can dynamically select the appropriate protocol according to the bandwidth and latency requirements of data transmission. For example, when transmitting large datasets, the RDMA protocol is preferred to reduce transmission latency, while the TCP / IP protocol is used to ensure compatibility when accessing small file resources. For sensitive data transmission that requires high security, the encrypted HTTPS protocol can be used. For high-speed transmission of large files, a private protocol based on UDP can be used to improve transmission efficiency.

[0045] In this network framework, the computing power pool and resource pool are managed independently through decoupling, eliminating the strong coupling between them. Specifically, when executing tasks, the computing power pool can dynamically select and establish connections with different resource pools based on actual needs to access the required data resources, rather than being limited to a pre-bound specific resource pool. This decoupling greatly improves the flexibility and efficiency of resource utilization, avoids the problem of duplicate storage caused by multiple computing power pools accessing the same data resource, effectively saves storage space, and reduces storage overhead. At the same time, it also allows the computing power pool to focus more on the execution of computing tasks, improving the overall system performance and stability.

[0046] The following is combined Figure 1The network framework described herein is used to describe the service provision method in the embodiments of this application.

[0047] Please see Figure 2 One embodiment of the service provision method in this application includes:

[0048] 201. In response to a tenant's access request, determine the target computing pool associated with the tenant based on the computing pool configuration information;

[0049] The method of this embodiment can be applied to a service providing device, which can be connected to... Figure 1 The network framework shown illustrates the connection of computing power pools. The service provider can obtain configuration information for both the resource pool and the computing power pool to control tenant access requests for data based on this information. A tenant refers to an entity that rents computing power and resource pool services; this can be an enterprise-level tenant (such as an internet company or research institution) or an individual developer. Tenant access requests typically include key information such as tenant identifier, task type (e.g., model training, data inference), required computing power scale (e.g., number of CPU cores, GPU memory size), and target data resource type (e.g., a specific version of a pre-trained model, labeled dataset).

[0050] The access request can be a resource access request initiated by the tenant on its device. For example, if the tenant starts an AI model training task on its device, and this task needs to obtain the corresponding network model and run it in the computing power pool, the service provider device can detect this access request and determine that the tenant needs to download the network model for model training in the computing power pool.

[0051] The configuration information of the computing power pool includes the basic attributes of the computing power pool and the tenant association rules. The basic attributes include the identifier of the computing power pool (such as a unique numerical code), computing power specifications (such as CPU model, number of GPUs, memory capacity), computing power type (such as general computing type, AI acceleration type), the current available computing power resource ratio (such as 70% of the total number of CPU cores, 30% of the GPU memory utilization rate), and geographical location (such as East China data center, North China data center), etc.

[0052] Tenant association rules define the binding relationship between tenants and computing power pools. For example, tenant A of an e-commerce company can be permanently associated with computing power pool 1 for daily data analysis tasks, while tenant A's temporary high-concurrency tasks can be dynamically associated with computing power pool 2; or, tenant A can be permanently associated with computing power pool 1, and tenant B can be permanently associated with computing power pool 2. Association rules can be dynamically adjusted based on dimensions such as tenant level (e.g., VIP tenants can be preferentially associated with high-specification computing power pools), task priority (e.g., urgent tasks are automatically matched with idle computing power pools), and historical usage records (e.g., priority is given to assigning commonly used computing power pool types to tenants).

[0053] The configuration information of the resource pool can include configuration details such as the resource pool's connection, data access, and relationships with other devices. These configurations help the service provider to accurately manage the interaction between the resource pool and the computing power pool.

[0054] The configuration information of the computing power pool can also include the access permissions of the computing power pool to the data of the resource pool, the relationship between the computing power pool and other devices, and other configurations. These configurations help the service provider to effectively control the access behavior of the computing power pool to the resource pool.

[0055] The association between a computing power pool and a tenant can be achieved by linking the pool's identification information to a tenant's account, or by linking the pool's identification information to a tenant's device identifier; this is not limited here. This association can be determined based on the creation relationship—that is, a computing power pool created by a tenant is associated with that tenant; or it can be determined based on the usage relationship—that is, a computing power pool used by a tenant through leasing is associated with that tenant; of course, this association can also be configured by personnel according to actual business needs.

[0056] 202. Determine the target data resource pool that the target computing power pool needs to connect to based on the resource pool configuration information; wherein, the resource pool configuration information is used to determine the association between the tenant and the data resource pool, or, the resource pool configuration information is used to determine the association between the computing power pool and the data resource pool; the resource pool configuration information can be adjusted by the tenant; the same data resource pool can be connected to different computing power pools;

[0057] After identifying the target computing power pool, the service provider can further determine the target data resource pool that the target computing power pool needs to connect to during task execution based on the resource pool's configuration information. The resource pool configuration information includes detailed information about various resource pools, such as storage type, storage location, and data access permissions. This information provides the service provider with a decision-making basis, enabling it to accurately match the target computing power pool with the corresponding resource pool.

[0058] The configuration information of the resource pool is used to determine the association between a tenant and a data resource pool, or the configuration information of the resource pool is used to determine the association between a computing power pool and a data resource pool. Based on this, the service provider can match the target data resource pool through two paths: if the resource pool configuration information defines the association rules between a tenant and a data resource pool, for example, if tenant B's model training task is associated with resource pool 3 which stores industry pre-trained models by default, then the service provider can directly lock the corresponding resource pool 3 based on the tenant identifier; if the configuration information defines the association rules between a computing power pool and a data resource pool, for example, if computing power pool 4 is bound to resource pool 4, then when the target computing power pool is computing power pool 4, resource pool 4 can be automatically matched as the target data resource pool.

[0059] The configuration information of resource pools can be adjusted by tenants, and the same data resource pool can be connected to different computing power pools. For example, tenant C originally configured its computing power pool 5 to be associated only with resource pool 5 that stores internal R&D datasets. When this tenant needs to carry out joint modeling tasks, it can independently modify the resource pool configuration information to add an association rule with resource pool 6 that stores publicly available datasets in the industry. This allows computing power pool 5 to automatically connect to resource pool 6 to obtain public data when performing joint modeling tasks. Moreover, resource pool 6 simultaneously supports connections and access from multiple different tenants' computing power pools, such as tenant D's computing power pool and tenant E's computing power pool, without having to copy a public dataset separately for each computing power pool. This simplifies the tenant's resource configuration process and achieves efficient sharing of data resources.

[0060] Optionally, the configuration information of the resource pool may also include configuration information for representing the attributes of the data resources stored in the resource pool. For example, if the data resources have shared or public attributes, it means that the data resources in the resource pool can be accessed and used by multiple tenants. In this case, the target computing power pool can be routed to the resource pool so that it can access the shared data resources in the resource pool.

[0061] 203. Establish a connection between the target computing power pool and the target data resource pool, utilize the target computing power pool to access the data resources stored in the target data resource pool based on the connection, and provide computing power services to the tenant;

[0062] The configuration information of the target data resource pool may include configuration information indicating its connection method. After the target computing power pool and the target data resource pool are determined, the service providing device can establish a connection between the target computing power pool and the target data resource pool based on the configuration information of the target data resource pool, such as the connection method configuration information. This connection establishment process may involve various network protocols and communication mechanisms to ensure the stability and security of data transmission. Once the connection is successfully established, the target computing power pool can access the data resources stored in the target data resource pool based on the connection, thereby meeting its task execution needs and providing the tenant with the required computing power services. This dynamic connection establishment method based on configuration information not only improves the flexibility of resource access but also enables the computing power pool to utilize storage resources more efficiently, avoiding waste caused by resource binding.

[0063] Therefore, in this embodiment, by decoupling the computing power pool from the resource pool, the computing power pool can dynamically select and connect to the target data resource pool to access the required data resources according to actual needs when executing tasks, instead of being limited to accessing only the resource pool it is bound to. Furthermore, since the same resource pool can be connected to different computing power pools, when multiple computing power pools need to access the same data resource, it is not necessary to store the same data resource in multiple resource pools; the data resource only needs to be stored in the common resource pool, avoiding duplicate storage of the same data resource and saving storage resources.

[0064] Secondly, resource pool permissions are configured on a tenant-by-tenant basis. The same computing power pool can adjust the resource pools it is connected to, realizing flexible scheduling of computing power and resources and avoiding single-point bottlenecks in computing power. That is, when a tenant's high-concurrency tasks are concentrated in a single computing power pool, the resource pools connected to that computing power pool can be dynamically switched to divert some data access requests to resource pool nodes with lower loads. At the same time, other idle computing power pools can be scheduled to connect to the original resource pool, thereby balancing the load pressure of computing power and storage.

[0065] At the same time, this decoupling approach also improves the flexibility and efficiency of resource utilization. The computing pool does not need to worry about which resource pool to select for data access; it only needs to establish a connection with the resource pool based on the connection configuration of the resource pool. This allows the computing pool to focus more on the execution of computing tasks, thereby improving the overall system performance and stability.

[0066] based on Figure 2 In one optional implementation of the embodiment shown, the configuration information of the resource pool may include storage configuration information, which includes tenant storage relationships and resource pool connection information. The tenant storage relationships are used to represent the association between the tenant and the resource pool.

[0067] In one application scenario, a tenant can upload file data to a resource pool, or read or modify data resources in the resource pool. Specifically, the target data resource pool associated with the tenant can be determined based on the tenant's storage relationship. A connection with the target data resource pool can be established based on the connection information of the target data resource pool, and the target storage system can be accessed based on the connection with the target data resource pool, such as uploading file data to the target storage system or reading / modifying data resources stored in the target storage system.

[0068] For example, such as Figure 3 This illustrates an application scenario where the file service is responsible for receiving requests from tenants to upload, download, and browse files, and for executing specific file transfer operations. It provides a unified resource operation interface to upper-layer applications. The management service is responsible for centrally managing the relationships and configuration information of all tenants, storage, and computing pools.

[0069] Management services and file services can be integrated into this service provider to jointly provide data access management and control services to the outside world.

[0070] External storage (such as external storage 1 and external storage 2, but not limited to these two) refers to a physically or logically independent storage system, serving as a resource pool used by the system and separate from the computing power pool. External storage can be any network storage device that supports standard protocols, such as enterprise-grade network attached storage (NAS), cloud storage services (such as object storage), or other distributed storage systems. Its stored content, including models and images, is a typical data resource in the field of AI research and development; of course, the stored data can also be data objects processed in other data processing technology fields. External storage 1 and external storage 2 can each function as a separate resource pool and provide data services independently, or they can work together as components of the same resource pool to provide data services; this is not limited here.

[0071] The "Storage Configuration Information" in the diagram serves as a storage space used for configuring storage resource pools. This includes storing the relationship between tenants and storage systems, as well as parameters required to access specific external storage, such as the storage system's protocol type, connection address, IP address, or domain name. Protocol types can include: NFS (Network File System), commonly used in Linux / Unix environments; CIFS / SMB (Common Internet File System / Server Message Block), a common network file-sharing protocol, commonly used in Windows environments and widely supported by other operating systems; and S3 (Simple Storage Service), an object storage protocol and the de facto standard for cloud object storage (such as AWS S3 and MinIO).

[0072] The following describes the operation process corresponding to the application scenario shown in the attached diagram:

[0073] 1. Configure storage: Configure external storage information for different tenants, establish the relationship between "tenant-storage system", dynamically associate tenants with external storage of different protocols, and decouple the storage system from the computing power pool;

[0074] 2. Record storage configuration: The management service records the storage relationships of different tenants and manages the association relationships, enabling the binding and unbinding of tenants from the storage system;

[0075] 3. File operations: Tenants can perform file operations on the platform, such as uploading resources like models and images to their storage system;

[0076] 4. Obtain tenant storage configuration: The file service requests the management service to read the configuration of the tenant storage relationship (i.e., the "tenant-storage system" association). Here, the file service and the management service are separated to isolate the configuration operation service from the file operation service.

[0077] 5. Retrieve tenant storage configuration information from records: After receiving the request, the management service reads the tenant storage relationship configuration from the "Storage Configuration Information" record and returns it to the file service;

[0078] 6. File Stream Operations: After obtaining the tenant's storage relationship configuration, the file service selects the corresponding protocol client to connect to the external storage system based on this configuration information. That is, according to the storage protocol specified in the configuration, it selects the appropriate software client to connect to the external storage to access it, such as for data upload, read, and modification. In this way, the file service does not need to be aware of the specific storage system's protocol type; it only needs to interface with a standard connection protocol to achieve integration with multiple storage systems.

[0079] Here, "protocol" refers to the network storage communication protocol, which is the rule and language for communication between the client and the storage server. "Client" refers to the software library, driver, or command-line tool that implements a specific storage protocol. For example, to access an NFS protocol-type storage system, the client needed is the operating system's `mount` command or the `libnfs` library; to access a CIFS protocol-type storage system, the client needed is the `smbclient` tool or the `cifs-utils` package; and to access an S3 protocol-type storage system, the client needed is the corresponding cloud vendor's SDK or a general S3 client library.

[0080] Therefore, through the configuration process described above, tenants can perform various data operations on the resource pool corresponding to the storage system without needing to concern themselves with the specific type and protocol details of the underlying storage system. This design greatly simplifies the tenant's operational process and improves the convenience and flexibility of data access. Tenants only need to perform simple configurations on the management platform to achieve seamless integration with different types of storage systems, thereby meeting diverse data storage and access needs.

[0081] At the same time, this decoupled design also brings significant system advantages. On the one hand, it allows the storage system to be expanded and upgraded independently of the computing pool, without worrying about compatibility issues with the computing pool. On the other hand, it also improves the system's reliability and fault tolerance. When a storage system fails, tenants can reconfigure to route data access requests to other working storage systems, thereby ensuring business continuity and data availability.

[0082] Furthermore, this design boasts excellent openness and scalability. It supports multiple network storage communication protocols, facilitating easy integration with various new storage systems and technologies. With the continuous development and innovation of storage technologies, tenants can easily integrate new storage systems into their existing data access control frameworks, thereby fully leveraging the advantages of new technologies and improving the overall performance and efficiency of the system.

[0083] based on Figure 2 In another optional implementation of the embodiment shown, the configuration information of the resource pool may include storage configuration information, which includes tenant storage relationships and connection information of the resource pool. The tenant storage relationships are used to represent the association between the tenant and the resource pool.

[0084] Furthermore, in one application scenario, when establishing a connection between a target computing power pool and a target data resource pool, the target data resource pool associated with the tenant can be determined based on the tenant storage relationship, and a connection between the target computing power pool and the target data resource pool can be established based on the connection information of the target data resource pool. Thus, when the target computing power pool needs to perform corresponding business services, it can obtain the corresponding data from the target data resource pool for processing based on the connection with the target data resource pool.

[0085] For example, such as Figure 4 This diagram illustrates an application scenario. The concepts of "storage configuration information," "management service," and "external storage" have already been explained previously and will not be repeated here. The following describes the operational flow corresponding to the application scenario shown in the attached diagram:

[0086] 1. Configure computing power pools: A single tenant can create multiple computing power pools. Computing power pools are associated with each other, and a computing power pool aggregates the computing power resources such as CPU, GPU, and memory of multiple physical machines.

[0087] 2. Record computing pool information: The computing pool information is the configuration information of the computing pool, which may include the relationship between the tenant and the computing pool and the relevant parameters of the computing pool. The management service can record the relationship and information between the tenant and the computing pool to realize the binding and unbinding functions of the tenant and the computing pool.

[0088] 3. Read storage configuration information: The management service reads the configuration of tenant storage relationships and the connection information of the storage system from the storage configuration information; the storage system can be used as a resource pool or a component of a resource pool for data storage and providing data access services;

[0089] 4. Distribute storage connection configuration: After a tenant binds to a computing power pool, the management service can perform a series of initialization actions on the computing power pool, including distributing the connection information of the storage system to all computing power pools;

[0090] 5. File operations of business services: Business services running on the computing power pool (such as AI training tasks) will connect to file storage through connection services such as storage CSI driver (Container Storage Interface Driver);

[0091] 6. Connection Service Reads Storage Configuration Information: The connection service reads the distributed storage configuration information and obtains the connection information of the external storage system. The connection service is a component running in the computing power pool (such as storage CSI driver, file system client), responsible for establishing the connection between business services and external storage systems.

[0092] 7. Use different protocols to connect to storage: The connection service uses the corresponding protocol to access the tenant's external storage system based on the storage configuration information.

[0093] The computing power pool can be dynamically bound to the storage system, thereby enabling resource sharing between tenant computing power pools. Tenants only need to upload one copy of the data to the external storage, and business services from multiple different computing power pools can share that file resource.

[0094] Therefore, through the above approach, tenants can flexibly manage and utilize computing and storage resources, eliminating the need to repeatedly store the same data in each computing pool, thus significantly saving storage space and costs. Simultaneously, this design also improves the efficiency and security of data access. Since the connectivity service is responsible for establishing connections between business services and external storage systems and using appropriate protocols based on the storage system's configuration information, this ensures the stability and security of data transmission.

[0095] Furthermore, when the storage system needs to be replaced or upgraded, tenants only need to update the storage system's configuration information, without modifying the code or configuration of business services, thereby reducing maintenance costs and risks. This decoupled design allows the system to better adapt to future business development and changes, providing effective support for business scenarios such as data access.

[0096] based on Figure 2 In another optional implementation of the embodiment shown, when the target computing power pool accesses the target data resource pool, the tenant's access permissions to the target data resource pool can be determined, thereby enabling the target computing power pool to access the target data resource pool according to the access permissions.

[0097] Access permissions can include read-write access and read-only access. Read-write access grants the user the right to read data resources stored in the target data resource pool, as well as the right to write new data resources or modify existing data resources in the target data resource pool. Read-only access grants the user only the right to read data resources stored in the target data resource pool, but does not allow writing or modifying data. This permission setting mechanism ensures the security and compliance of data access, preventing unauthorized data access and modification.

[0098] Specifically, if a tenant is the creator of the target data resource pool, then the tenant has read and write access permissions to the target data resource pool; if a tenant has access permissions to the target data resource pool but is not the creator of the target data resource pool, then the tenant has read-only access permissions. In other words, the creator of the data resource has the highest level of data control, and can independently decide to update, delete, or authorize other tenants to access the data; while non-creation tenants can only read the data within the authorized scope and cannot make any changes to the data content, effectively avoiding the risk of accidental modification or malicious tampering of the data.

[0099] For example, such as Figure 5 This diagram illustrates an application scenario. The concepts of "storage configuration information," "management service," and "external storage" have already been explained previously and will not be repeated here. The following describes the operational flow corresponding to the application scenario shown in the attached diagram:

[0100] 1. Associating the computing power pool with tenants: A tenant is a tenant who rents and uses a computing power pool. The administrator, as the owner of the computing power pool, has the right to allocate the computing power pool to tenants and grant tenants certain usage permissions to the computing power pool. After being assigned usage permissions, tenants can use the computing power pool to access data resources stored in the private storage system and the shared storage system of the administrator.

[0101] 2. Record association relationships: After receiving the association request between the computing power pool and the tenant from the administrator, the management service will record the association relationship between the computing power pool and the tenant, and realize the dynamic association / unassociation of the computing power pool with the tenant through the management service;

[0102] 3. Read the storage configuration information of the administrator and tenant: The management service obtains the storage configuration information of the administrator and the tenant from the record information; the storage configuration information of the administrator is the configuration information of the storage system associated with the administrator, and the storage configuration information of the tenant is the configuration information of the storage system associated with the tenant.

[0103] 4. Distribute storage configuration: After associating tenants, a series of tenant namespace initializations can be performed in the computing power pool. During this process, the configuration information of the administrator's shared storage system and the configuration information of the tenant's personal storage system will be distributed to the tenant's namespace. Each tenant has an independent namespace, thereby achieving space isolation between tenants.

[0104] 5. Business service file operations: For tenants with assigned permissions, the business services running in the computing power pool associated with that tenant can read data resources from the tenant's private storage system and write data to the tenant's private storage system. For shared storage systems, business services can only read data resources from the shared storage system and do not have write permissions.

[0105] 6. Read connection information: The connection service reads the configuration information of the corresponding storage system according to the configuration information;

[0106] 7. Connecting storage using different protocols: The connection service connects to the corresponding storage system based on the storage system's configuration information, such as connecting to a shared storage system managed by the administrator or connecting to a tenant's private storage system, and performs corresponding data access operations, such as data reading and writing operations.

[0107] based on Figure 5 In the scenario shown, when establishing a connection with the target data resource pool based on the connection information of the target storage system, the target data resource pool can be either a tenant's private data resource pool or a shared data resource pool. A private data resource pool supports access by tenants with corresponding access permissions, while a shared data resource pool supports access by any tenant.

[0108] The shared data resource pool can be created by tenants of any level or with any identity. For example, it can be created by an administrator or by a regular tenant without administrator status. There are no restrictions here.

[0109] The owner of a private data resource pool has access rights to the private data resource pool. The owner can set the private data resource pool to be accessible only to the owner, or the owner can assign access rights to the private data resource pool to other tenants, such as assigning read-only permissions to other tenants.

[0110] Specifically, connection information for the shared data resource pool and the private data resource pool can be distributed to the tenant's namespace. Then, a connection can be established between the computing power pool and the shared data resource pool based on the connection information of the shared data resource pool, or between the computing power pool and the private data resource pool based on the connection information of the private data resource pool.

[0111] The private data resource pool provides read and write permissions to the computing power pools of tenants with read and write access to that resource pool. The shared data resource pool provides read-only permissions to the computing power pools of a wider range of tenants. Therefore, when a tenant's computing power pool with the appropriate permissions accesses its corresponding private data resource pool, it accesses the data resources stored in the private data resource pool using the read and / or write operation methods set by the read and write permissions. When a tenant's computing power pool accesses the shared data resource pool, it accesses the data resources stored in the shared data resource pool using the read-only operation method set by the read-only permissions.

[0112] For example, Figure 6 The diagram illustrates shared and dedicated computing power pool scenarios. In these scenarios, the creator of the computing power pool has usage rights, including access to its computing power and privately stored file resources. For example, in a shared computing power pool scenario, the administrator has usage and management rights for the computing power pool group they create. The computing power pool group belongs to the administrator, who can allocate computing power pools within the group to other tenants.

[0113] Shared computing power pools can dynamically mount shared data resource pools and tenants' private data resource pools, allowing different tenants to access data from their respective pools based on their permissions. In a shared computing power pool scenario, computing power pools allocated by the administrator to other tenants can only be operated according to the permissions granted to them. If access is granted to a shared data resource pool, it can only read data resources within the pool using read-only permissions, preventing write or modification operations. This ensures the security and integrity of shared data, preventing unauthorized data tampering.

[0114] For dedicated computing power pools, the creator of the pool has full access rights. The creator can utilize the pool's computing power and freely access their private storage resources, performing read, write, and modify operations on the data stored in their private data pool using the read and write permissions set accordingly. This design not only meets the diverse needs of different tenants in different scenarios but also ensures data access security and compliance through clear permission settings, making the entire system more efficient and orderly in data management and utilization.

[0115] In the above scenario, the computing power pool and resource pool are decoupled. Different tenants can have their own private data resource pools, and resources within the same tenant's dedicated computing power pools can be shared. For example, a tenant creates computing power pool A and computing power pool B. Both computing power pool A and computing power pool B can be linked to the tenant's dedicated private data resource pool, sharing the data resources within it. When a tenant needs to use computing power pool A or computing power pool B for business processing, regardless of which computing power pool is selected, they can directly access the same data in their private data resource pool without needing to repeatedly copy or transfer data between different computing power pools. This not only improves the convenience of data use but also avoids various problems that may arise due to data inconsistency.

[0116] Each data resource has a corresponding attribute tag, which indicates whether the data resource is private or shared. If the attribute tag of the data resource to be accessed by the resource access request indicates a private attribute, then the data resource belongs to the private data resource pool; if the attribute tag of the data resource indicates a shared attribute, then the data resource belongs to the shared data resource pool.

[0117] This attribute label can be adjusted to allow data resources to switch between private and shared attributes based on actual business needs. In other words, the attribute label of a data resource can be modified to control whether the data resource can be shared or accessed only privately. For example, when a request to change the attribute of a data resource is received, the current attribute label of the data resource can be changed to the attribute label indicated by the request.

[0118] For example, such as Figure 7 This embodiment illustrates another application scenario where access permissions for data resources are managed by upper-layer business logic. The private or shared attributes of data resources are controlled by labeling their metadata. Tenants' access permissions to the data resource pool control their visibility of the data resources. The advantage of this design is that when the attributes of data resources change, such as switching between "private" and "public," only the metadata labels of the data resources need to be changed, without requiring data migration. That is, there is no need to migrate data resources from a shared data resource pool to a private data resource pool, or vice versa, significantly reducing the operational burden of data management and making it more business-friendly.

[0119] In this scenario, administrators can upload data resources such as models with shared and private attributes. The management service can generate and store metadata for these data resources, tagged with shared and private attributes. Shared attribute tags for multiple data resources can be stored as shared resource information, and private attribute tags for multiple data resources can be stored as private resource information. This allows the computing power pool to determine whether to route data access requests to the shared or private data resource pool based on the attribute tags of the data resources it needs to access. Furthermore, data resources can be uploaded to the shared data resource pool for authorized tenants to share and access.

[0120] Tenants upload and use private data resources. Tenants also have the right to upload and use private data resources such as models. These data resources can be uploaded to the tenant's own data resource pool, which is inaccessible to others.

[0121] Tenants use shared resources. Once the administrator shares the computing power pool with a tenant, the shared resources become visible to the tenant, who can use them as if they were private resources. However, the tenant cannot perform write operations on the shared resources. This is controlled by using account permissions when connecting the service to the shared data resource pool. Specifically, the tenant's account is identified to determine whether they have write permissions. If the account permission is read-only (e.g., if it is a tenant's account), only data reading operations are allowed. This ensures that the shared resources are not modified arbitrarily and maintains data stability and security.

[0122] In some optional implementations, the data resource pool supports partitioning storage space on a tenant-by-tenant basis. For example, a private data resource pool and a shared data resource pool of the same tenant A can share the same storage space. Furthermore, for any other tenant, such as tenant B, regardless of their access permissions (read / write / read-only), when tenant B accesses tenant A's private data resource pool or shared data resource pool, they are accessing the same storage space. However, the data resources visible to tenant B in that storage space can be determined based on tenant B's access permissions.

[0123] For example, tenant A's private data resource pool and shared data resource pool share the same storage space. This storage space is internally divided into a "private data area" and a "shared data area," but there is no need for physical partitioning; logical distinction is made only by the attribute tags of the data resources. When tenant B attempts to access this storage space, the system first reads tenant B's permission configuration: if tenant B is only granted permission to access tenant A's shared data resource pool, the system will filter out data resources with the attribute tag "shared," only show tenant B access to these resources, and restrict their operations to read-only (such as viewing or downloading copies of data resources); if tenant B is granted permission to access tenant A's private data resource pool (such as if tenant A actively authorized it), the system will filter out data resources with the attribute tag "private" that are open to tenant B. In this case, tenant B's operation permissions are determined by the authorized content—if it is read-write permission, they can modify, add, or delete this part of the private data resources; if it is read-only permission, they can only view or download.

[0124] This logical isolation method based on attribute tags and permission configuration saves the cost of physical partitioning of storage space, and ensures the security and accuracy of data access by different tenants through dynamic permission verification, avoiding resource waste or permission confusion caused by physical storage partitioning.

[0125] Meanwhile, private data resources uploaded by tenants to the private data resource pool are inaccessible to other tenants and administrators, fully protecting tenant privacy and data security. This granular access control mechanism enables different tenants to accurately and securely access the required data resources in different scenarios, meeting diverse business needs while ensuring compliance and security of data access. Moreover, when the attributes of data resources need to be changed, only their metadata tags need to be modified, eliminating the need for cumbersome data migration operations. This greatly simplifies the data management process, improves the system's flexibility and scalability, and provides effective assurance for the efficient operation of the entire service delivery method.

[0126] In this embodiment, the resource pool and the computing power pool can be dynamically associated based on access permissions. The computing power pool can accurately access the data resources in the corresponding data resource pool within the resource pool according to the granted access permissions. This dynamic association mechanism breaks the limitation of the traditional model where computing power and storage are tightly bound together, and realizes a more flexible and efficient configuration and utilization of computing power resources and storage resources.

[0127] In practical applications, this mechanism can smoothly facilitate both on-demand access to public data resources by multiple tenants in a shared scenario and comprehensive operation of private data resources by tenants in a dedicated scenario. Furthermore, the permission management method based on data resource attribute tags further refines the granularity of data access, ensuring flexible conversion between shared and private data without the need for large-scale data migration, thus greatly improving the convenience and efficiency of data management.

[0128] The following is combined Figure 1 The network framework described herein describes the tenant configuration management method in the embodiments of this application:

[0129] Please see Figure 8 One embodiment of the tenant configuration management method in this application includes:

[0130] 801. In response to a tenant's computing pool configuration command, create or adjust the configuration information of a computing pool; the configuration information of the computing pool is used to determine the target computing pool associated with the tenant;

[0131] 802. In response to a tenant's resource pool configuration command, create or adjust the configuration information of a resource pool; wherein, the configuration information of the resource pool is used to determine the association between the tenant and the data resource pool, or to determine the association between the computing power pool and the data resource pool; the computing power pool and the data resource pool are decoupled so that the same data resource pool can be connected to different computing power pools.

[0132] The method in this embodiment can be applied to a tenant's configuration management module. This module can be a functional component used by the tenant to manage and configure the association between its own computing power pool and data resource pool, or it can be a core module on the system side that provides configuration management services to the tenant. This module receives computing power pool configuration commands and resource pool configuration commands initiated by the tenant to achieve dynamic scheduling and association management of the tenant's computing power resources and storage resources.

[0133] This embodiment achieves decoupled management of computing resources and storage resources by responding separately to computing pool configuration commands and resource pool configuration commands. Creating or adjusting the configuration information of a computing pool can involve adding one or more associations between computing pools and tenants, or adjusting existing associations between computing pools and tenants. Therefore, when a tenant initiates a computing pool configuration command, the system can create a new computing pool or adjust the configuration information of an existing computing pool according to the command requirements, such as adding an association between a computing pool and a tenant, or adjusting an existing association between a computing pool and a tenant.

[0134] The so-called "adjustment" can be modifying or deleting one or more associations between computing power pools and tenants.

[0135] For example, to meet business expansion needs, tenant A initiates a computing pool configuration command, requesting the addition of computing pool C. Upon receiving this command, the system first verifies tenant A's account permissions (e.g., whether they are qualified to create a computing pool). If correct, it automatically allocates computing resources (including quotas for hardware resources such as CPU, GPU, and memory) and generates a unique identifier for computing pool C. Simultaneously, the association between computing pool C and tenant A is written to the configuration database, completing the new configuration of computing pool C. Subsequently, tenant A can quickly access the computing power of computing pool C using this identifier, without needing to repeatedly apply for hardware resources, significantly shortening the business deployment cycle.

[0136] Subsequently, the association between computing pool C and tenants can also be adjusted. For example, when tenant A's business scale shrinks and computing pool C is no longer needed, a computing pool configuration command can be initiated to request the removal of the association with computing pool C. After receiving the command, the system will first check whether computing pool C is still running tasks: if there are unfinished tasks, the system will prompt tenant A to wait until the tasks are completed before proceeding, or allow tenant A to choose to forcibly remove the association (in this case, unfinished tasks will be terminated and breakpoints will be saved); if computing pool C is idle, the system will directly delete the association record between tenant A and computing pool C from the configuration database, while retaining the physical resources of computing pool C (which can be reused by other tenants). This adjustment method ensures business continuity and improves the utilization rate of computing resources.

[0137] Resource pool configuration commands focus on adjusting the relationships between tenants and data resource pools, and between computing power pools and data resource pools. For example, they can attach a tenant's private data resource pool to its dedicated computing power pool, or dynamically associate a shared computing power pool with a shared data resource pool. This decoupled design allows the same data resource pool to be connected to multiple different computing power pools simultaneously, avoiding the problems of resource idleness or redundant configuration in the traditional binding model, and significantly improving resource reuse and configuration flexibility.

[0138] For example, a company's marketing department and R&D department each use dedicated computing power pools to process business, but both need to access the company's shared user behavior dataset (stored in a shared data resource pool). Through resource pool configuration commands, the system can simultaneously mount the shared data resource pool to the computing power pools of both departments, eliminating the need to copy the dataset separately for each department. This saves storage costs and ensures data consistency.

[0139] Similarly, creating or adjusting the configuration information of a resource pool can involve adding one or more associations between a tenant and a data resource pool, or between a computing power pool and a data resource pool, or adjusting existing associations. "Adjustment" can refer to modifying or deleting one or more associations between a tenant and a data resource pool, or between a computing power pool and a data resource pool.

[0140] For example, when a tenant's R&D department completes a phase of a project and no longer needs to access the dataset in the shared data resource pool, it can initiate a resource pool configuration command to request the removal of the association between its dedicated computing pool and the shared data resource pool. Upon receiving the command, the system first verifies the R&D department's permissions (e.g., whether it is the administrator of the computing pool). If confirmed, it directly deletes the association record between the computing pool and the shared data resource pool from the configuration database. At this point, the computing pool will no longer be able to access the data in the shared data resource pool, but the shared data resource pool itself can still be accessed normally by computing pools from other departments. If the R&D department subsequently needs to re-associate the shared data resource pool due to new project requirements, it only needs to initiate the resource pool configuration command again, and the system can quickly restore the association. The entire process does not require modification of the data resource pool's physical storage or permission settings, greatly simplifying the operation process.

[0141] In one optional implementation, the attribute tags of the data resources can also be adjusted in response to the tenant's data resource configuration instructions. These attribute tags indicate whether the data resource is private or shared. If the attribute tag indicates a private attribute, the data resource belongs to a private data resource pool; if the attribute tag indicates a shared attribute, the data resource belongs to a shared data resource pool.

[0142] Furthermore, this attribute tag can be adjusted, making it easy to switch between private and shared attributes for data resources. Private data resource pools support access by tenants with corresponding access permissions, while shared data resource pools support access by any tenant.

[0143] The concepts of "private attribute", "shared attribute", "private data resource pool" and "shared data resource pool" mentioned above are similar to the corresponding concepts in the previous embodiment. Therefore, the explanation of the above concepts can be found in the explanation of the corresponding content above, and will not be repeated here.

[0144] Therefore, in this embodiment, the tenant configuration management module constructs a data access and computing power scheduling system centered on the tenant and based on resource decoupling by flexibly adjusting the computing power pool configuration, resource pool configuration, and data resource attribute tags. This system not only ensures the security of tenant's private data and the independence of dedicated computing power, but also achieves efficient resource reuse across tenants and departments through the logical management of shared resources, effectively solving the problems of operational complexity and high costs caused by resource binding in the traditional model.

[0145] For example, when multiple business lines of an enterprise tenant need to collaboratively process data, the administrator can use a single resource pool configuration command to simultaneously associate the shared data resource pool with the computing power pools of each business line. Each business line obtains a consistent data source under read-only permissions, avoiding data redundancy and version inconsistencies. Furthermore, when a business line needs to convert its internal model to a shared resource, it only needs to modify the model's attribute tags, and the system will automatically include it in the shared resource management scope, without requiring additional data migration operations. This lightweight configuration mode not only reduces the management burden on operations personnel but also provides more flexible resource support for tenant business innovation, helping tenants quickly respond to needs in dynamically changing business scenarios and improve overall operational efficiency.

[0146] The following is combined Figure 2 and Figure 8 The illustrated embodiments describe a reasoning service generation system according to the present application. One embodiment of the reasoning service generation system in this application includes:

[0147] The inference service generation system includes:

[0148] The system comprises a control module, a computing power pool module, and a data resource pool module. The computing power pool module includes at least one computing power pool, and the data resource pool module includes at least one data resource pool. The computing power pool module and the data resource pool module are decoupled to enable the same data resource pool to be connected to different computing power pools.

[0149] The control module is used to execute Figure 2 The method described in the illustrated embodiments and its various alternative implementations is used to control the computing power pool to connect to the data resource pool in response to a tenant's access request, access the data resources provided by the data resource pool, and provide computing power services to the tenant;

[0150] The control module is also used to perform Figure 8 The method described in the illustrated embodiments and various alternative implementations is used to determine the computing pool associated with a tenant in response to a tenant's computing pool configuration instruction; and also to determine the association between a tenant and a data resource pool, or the association between a computing pool and a data resource pool, in response to a tenant's resource pool configuration instruction.

[0151] In this embodiment, the operations performed by each module in the inference service generation system are the same as described above. Figure 2 and Figure 8 The embodiments shown and their respective alternative implementations are similar to those described, and will not be repeated here.

[0152] The following is combined Figure 2 The illustrated embodiments describe the service providing apparatus according to the present application. One embodiment of the service providing apparatus in this application includes:

[0153] The computing power pool association module is used to respond to a tenant's access request and determine the target computing power pool associated with the tenant based on the computing power pool's configuration information.

[0154] The data resource pool association module is used to determine the target data resource pool that the target computing power pool needs to connect to based on the resource pool configuration information; wherein, the resource pool configuration information is used to determine the association relationship between the tenant and the data resource pool, or to determine the association relationship between the computing power pool and the data resource pool; the resource pool configuration information can be adjusted; the same data resource pool can be connected to different computing power pools;

[0155] The connection module is used to establish a connection between the target computing power pool and the target data resource pool, utilize the target computing power pool to access the data resources stored in the target data resource pool based on the connection, and provide computing power services to the tenant.

[0156] In some alternative implementations, the service providing apparatus further includes:

[0157] The determination module is used to determine the tenant's access permissions to the target data resource pool, so that the target computing power pool can access the target data resource pool according to the access permissions;

[0158] The access permissions include read-write access permissions and read-only access permissions; if the tenant is the creation tenant of the target data resource pool, then the tenant has read-write access permissions to the target data resource pool; if the tenant has access permissions to the target data resource pool, but the tenant is not the creation tenant of the target data resource pool, then the tenant has read-only access permissions.

[0159] In some optional implementations, the data resource pool includes a private data resource pool and a shared data resource pool. The private data resource pool supports access by tenants with corresponding access permissions, while the shared data resource pool supports access by any tenant.

[0160] In some optional implementations, the data resource has attribute tags, which are used to indicate whether the data resource has private or shared attributes;

[0161] If the attribute label of the data resource indicates a private attribute, then the data resource belongs to the private data resource pool; if the attribute label of the data resource indicates a shared attribute, then the data resource belongs to the shared data resource pool; the attribute label can be adjusted to allow the data resource to switch between private and shared attributes.

[0162] In some optional implementations, the configuration information of the resource pool includes connection information for the target data resource pool;

[0163] The connection module is specifically used to establish a connection between the target computing power pool and the target data resource pool based on the connection information of the target data resource pool.

[0164] In this embodiment, the operations performed by each module in the service providing device are the same as those described above. Figure 2 The embodiments shown and their various alternative implementations are similar to those described herein, and will not be repeated here.

[0165] The following is combined Figure 8 The illustrated embodiment describes the tenant configuration management module in this application. One embodiment of the tenant configuration management module in this application includes:

[0166] The computing pool configuration module is used to create or adjust the configuration information of computing pools in response to the tenant's computing pool configuration instructions.

[0167] The data resource pool configuration module is used to respond to tenants' resource pool configuration commands and create or adjust the configuration information of resource pools.

[0168] The configuration information of the computing power pool is used to determine the target computing power pool associated with the tenant; the configuration information of the resource pool is used to determine the association between the tenant and the data resource pool, or the association between the computing power pool and the data resource pool; the computing power pool and the data resource pool are decoupled so that the same data resource pool can be connected to different computing power pools.

[0169] In some optional implementations, the tenant configuration management module further includes:

[0170] The data resource configuration module is used to adjust the attribute tags of the data resource in response to the tenant's data resource configuration instructions;

[0171] The attribute tags are used to indicate whether the data resource is a private attribute or a shared attribute. If the attribute tag of the data resource indicates a private attribute, the data resource belongs to a private data resource pool; if the attribute tag of the data resource indicates a shared attribute, the data resource belongs to a shared data resource pool. The attribute tags can be adjusted to allow the data resource to switch between private and shared attributes.

[0172] The private data resource pool supports access by tenants with corresponding access permissions, while the shared data resource pool supports access by any tenant.

[0173] In this embodiment, the operations performed by each module in the tenant configuration management module are the same as those described above. Figure 8 The embodiments shown and their various alternative implementations are similar to those described herein, and will not be repeated here.

[0174] The computer device in the embodiments of this application is described below. Please refer to [link / reference]. Figure 9 One embodiment of the computer device in this application includes:

[0175] The computer device 900 may include one or more central processing units (CPUs) 901 and a memory 905, in which one or more applications or data are stored.

[0176] The memory 905 can be volatile or persistent storage. The program stored in the memory 905 can include one or more modules, each module including a series of instruction operations on the computer device. Furthermore, the central processing unit 901 can be configured to communicate with the memory 905 and execute the series of instruction operations stored in the memory 905 on the computer device 900.

[0177] The computer device 900 may also include one or more power supplies 902, one or more wired or wireless network interfaces 903, one or more input / output interfaces 904, and / or one or more operating systems, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0178] The central processing unit 901 can perform the aforementioned... Figure 2 The operations performed by the service providing device in the illustrated embodiments and their various implementations, or the execution of... Figure 8 The specific operations performed by the tenant configuration management module in the illustrated embodiment and its various implementations will not be described in detail here.

[0179] This application also provides a computer storage medium, one embodiment of which includes: the computer storage medium storing instructions, which, when executed on a computer, cause the computer to perform the aforementioned... Figure 2 The operations performed by the service providing device in the illustrated embodiments and their various implementations, or the execution of... Figure 8 The operations performed by the tenant configuration management module in the illustrated embodiment and its various implementations.

[0180] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0181] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0182] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0183] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0184] If the integrated unit is implemented as a software functional 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 this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A service provision method, characterized in that, The method includes: In response to a tenant's access request, the target computing pool associated with the tenant is determined based on the computing pool's configuration information. The target data resource pool to be connected to by the target computing power pool is determined based on the configuration information of the resource pool; wherein, the configuration information of the resource pool is used to determine the association between the tenant and the data resource pool, or, the configuration information of the resource pool is used to determine the association between the computing power pool and the data resource pool; the configuration information of the resource pool can be adjusted by the tenant; the same data resource pool can be connected to different computing power pools; Establish a connection between the target computing power pool and the target data resource pool, utilize the target computing power pool to access the data resources stored in the target data resource pool based on the connection, and provide computing power services to the tenant.

2. The method according to claim 1, characterized in that, The method further includes: Determine the tenant's access permissions to the target data resource pool, and enable the target computing power pool to access the target data resource pool according to the access permissions; The access permissions include read-write access permissions and read-only access permissions; if the tenant is the creation tenant of the target data resource pool, then the tenant has read-write access permissions to the target data resource pool; if the tenant has access permissions to the target data resource pool, but the tenant is not the creation tenant of the target data resource pool, then the tenant has read-only access permissions.

3. The method according to claim 1 or 2, characterized in that, The data resource pool includes a private data resource pool and a shared data resource pool. The private data resource pool supports access by tenants with corresponding access permissions, while the shared data resource pool supports access by any tenant.

4. The method according to claim 3, characterized in that, The data resource has attribute tags, which are used to indicate whether the data resource has private attributes or shared attributes; If the attribute label of the data resource indicates a private attribute, then the data resource belongs to the private data resource pool; if the attribute label of the data resource indicates a shared attribute, then the data resource belongs to the shared data resource pool; the attribute label can be adjusted to allow the data resource to switch between private and shared attributes.

5. The method according to claim 1 or 2, characterized in that, The configuration information of the resource pool includes the connection information of the target data resource pool; The establishment of the connection between the target computing power pool and the target data resource pool includes: The connection between the target computing power pool and the target data resource pool is established based on the connection information of the target data resource pool.

6. A tenant configuration management method, characterized in that, The method includes: In response to tenants' computing pool configuration commands, create or adjust the configuration information of computing pools; In response to tenants' resource pool configuration instructions, create or adjust the configuration information of resource pools; The configuration information of the computing power pool is used to determine the target computing power pool associated with the tenant; the configuration information of the resource pool is used to determine the association between the tenant and the data resource pool, or the association between the computing power pool and the data resource pool; the computing power pool and the data resource pool are decoupled so that the same data resource pool can be connected to different computing power pools.

7. The method according to claim 6, characterized in that, The method further includes: In response to the tenant's data resource configuration instructions, adjust the attribute labels of the data resource; The attribute tags are used to indicate whether the data resource is a private attribute or a shared attribute. If the attribute tag of the data resource indicates a private attribute, the data resource belongs to a private data resource pool; if the attribute tag of the data resource indicates a shared attribute, the data resource belongs to a shared data resource pool. The attribute tags can be adjusted to allow the data resource to switch between private and shared attributes. The private data resource pool supports access by tenants with corresponding access permissions, while the shared data resource pool supports access by any tenant.

8. A system for generating inference services, characterized in that, The system includes: The system comprises a control module, a computing power pool module, and a data resource pool module. The computing power pool module includes at least one computing power pool, and the data resource pool module includes at least one data resource pool. The computing power pool module and the data resource pool module are decoupled to enable the same data resource pool to be connected to different computing power pools. The control module is configured to execute the method of any one of claims 1 to 5, in response to a tenant's access request, to control the computing power pool to connect to the data resource pool, access the data resources provided by the data resource pool, and provide computing power services to the tenant; The control module is also configured to execute the method of claim 6 or 7 to determine the computing pool associated with the tenant in response to the tenant's computing pool configuration instruction; and to determine the association between the tenant and the data resource pool, or the association between the computing pool and the data resource pool, in response to the tenant's resource pool configuration instruction.

9. A service providing device, characterized in that, include: The computing power pool association module is used to respond to a tenant's access request and determine the target computing power pool associated with the tenant based on the computing power pool's configuration information. The data resource pool association module is used to determine the target data resource pool that the target computing power pool needs to connect to based on the resource pool configuration information; wherein, the resource pool configuration information is used to determine the association relationship between the tenant and the data resource pool, or to determine the association relationship between the computing power pool and the data resource pool; the resource pool configuration information can be adjusted; the same data resource pool can be connected to different computing power pools; The connection module is used to establish a connection between the target computing power pool and the target data resource pool, utilize the target computing power pool to access the data resources stored in the target data resource pool based on the connection, and provide computing power services to the tenant.

10. A tenant configuration management module, characterized in that, include: The computing pool configuration module is used to create or adjust the configuration information of computing pools in response to the tenant's computing pool configuration instructions. The data resource pool configuration module is used to respond to tenants' resource pool configuration commands and create or adjust the configuration information of resource pools. The configuration information of the computing power pool is used to determine the target computing power pool associated with the tenant; the configuration information of the resource pool is used to determine the association between the tenant and the data resource pool, or the association between the computing power pool and the data resource pool; the computing power pool and the data resource pool are decoupled so that the same data resource pool can be connected to different computing power pools.

11. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 7.

12. A computer storage medium, characterized in that, The computer storage medium stores instructions that, when executed on the computer, cause the computer to perform the method as described in any one of claims 1 to 7.

Images