Distributed AI-based data management method and terminal

Abstract

The invention discloses a distributed AI-based data management method and a terminal, a plurality of distributed AI nodes are set, each distributed AI node comprises an IP resource pool and a data feature, the IP resource pool comprises IP resource information of a host added to the distributed AI node, and the data feature comprises file resource information contained in each IP resource in the IP resource pool; receiving a data management request of a client, wherein the data management request comprises a source IP to which the client belongs; determining an operation object matched with the data management request according to the source IP, each distributed AI node and the corresponding IP resource pool and data characteristics, and executing an operation corresponding to the data management request according to the operation object; by means of the intelligent property of the AI, safe and reasonable allocation of resources is achieved through the AI technology, a distributed data source warehouse is formed by distributed AI nodes, the safety and reliability of file data management are guaranteed, and the method can be suitable for multi-scene file data management such as company-level file data management.

Description

Technical Field

[0001] The present invention relates to the field of data management, and in particular to a distributed AI-based data management method and terminal. Background Art

[0002] In the prior art, in order to download user data, data mining of the entire network can usually be performed through Thunder Cloud, P2P acceleration, and blockchain technology. Through the entire network data mining, the mirror user with the same data resource as the data to be downloaded is automatically matched for downloading. However, once the user downloads, the above-mentioned entire network data mining only allows a virtual location to exist, which is unique, and the broadband of the user's computer will be occupied when it is silent. In addition, the above-mentioned entire network data mining usually serves the general public, not enterprise-level users. The downloaded data source is uncertain and the security of the data source cannot be guaranteed. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a data management method and terminal based on distributed AI, which can be applicable to enterprise-level file data management and ensure the security and reliability of file data management.

[0004] In order to solve the above technical problems, a technical solution adopted by the present invention is:

[0005] A data management method based on distributed AI, comprising the steps of:

[0006] S1. Set up multiple distributed AI nodes, each distributed AI node includes an IP resource pool and data features, the IP resource pool includes IP resource information of hosts added to the distributed AI node, and the data features include file resource information included under each IP resource in the IP resource pool;

[0007] S2. Receive a data management request from a client, where the data management request includes a source IP address to which the client belongs;

[0008] S3. Determine the operation object that matches the data management request based on the source IP and each distributed AI node and its corresponding IP resource pool and data characteristics, and perform the operation corresponding to the data management request based on the operation object.

[0009] In order to solve the above technical problems, another technical solution adopted by the present invention is:

[0010] A distributed AI-based data management terminal comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, each step in the above-mentioned distributed AI-based data management method is implemented.

[0011] The beneficial effects of the present invention are: setting up multiple distributed AI nodes, each distributed AI node includes IP resource information of each host joining the node and file resource information contained under each IP resource. When a data management request from a client is received, the source IP to which the client belongs is determined, and an operation object matching the data management request is determined based on the source IP and each distributed AI node and its corresponding IP resource pool and data characteristics. The operation corresponding to the data management request is performed according to the operation object. Only clients joining the distributed AI nodes can perform data management within the range of hosts included in each distributed AI node. When performing data management, the IP resource information of each host and the file resource information contained under each IP resource are comprehensively considered. With the help of the intelligent nature of AI, AI technology is used to achieve safe and reasonable allocation of resources. A distributed data source warehouse is formed by the distributed AI nodes, which ensures the security and reliability of file data management and can be applied to multiple scenarios, such as company-level file data management. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A flowchart of a distributed AI-based data management method according to an embodiment of the present invention;

[0013] Figure 2 This is a schematic diagram of the structure of a distributed AI-based data management terminal according to an embodiment of the present invention;

[0014] Figure 3 This is a structural diagram of the IP resource pool and data characteristics of the distributed AI node in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in conjunction with the implementation modes and the accompanying drawings.

[0016] Please refer to Figure 1 , a data management method based on distributed AI, comprising the steps of:

[0017] S1. Set up multiple distributed AI nodes, each distributed AI node includes an IP resource pool and data features, the IP resource pool includes IP resource information of hosts added to the distributed AI node, and the data features include file resource information included under each IP resource in the IP resource pool;

[0018] S2. Receive a data management request from a client, where the data management request includes a source IP address to which the client belongs;

[0019] S3. Determine the operation object that matches the data management request based on the source IP and each distributed AI node and its corresponding IP resource pool and data characteristics, and perform the operation corresponding to the data management request based on the operation object.

[0020] From the above description, it can be seen that the beneficial effects of the present invention are: setting up multiple distributed AI nodes, each distributed AI node includes the IP resource information of each host joining the node and the file resource information contained under each IP resource. When a data management request from a client is received, the source IP to which the client belongs is determined, and the operation object matching the data management request is determined based on the source IP and each distributed AI node and its corresponding IP resource pool and data characteristics. The operation corresponding to the data management request is performed according to the operation object. Only clients joining the distributed AI nodes can perform data management within the range of hosts included in each distributed AI node, and when performing data management, the IP resource information of each host and the file resource information contained under each IP resource are comprehensively considered. With the help of the intelligent nature of AI, AI technology is used to achieve safe and reasonable allocation of resources. A distributed data source warehouse is formed by the distributed AI nodes, which ensures the security and reliability of file data management and can be applied to multiple scenarios, such as company-level file data management.

[0021] Furthermore, the IP resource information includes bandwidth usage, system status, and the previous distributed AI node identifier and the next distributed AI node identifier of each IP resource.

[0022] The S3 includes:

[0023] S31. Determine the source distributed AI node to which the source IP belongs according to the source IP;

[0024] S32. Determine the operation object that matches the data management request based on the data management request, the bandwidth usage of each IP resource of the host that joins the source distributed AI node, the system status, the previous distributed AI node identifier and the next distributed AI node identifier of the source distributed AI node, and the file resource information contained in each IP resource in the IP resource pool of the distributed AI node involved, and perform the operation corresponding to the data management request based on the operation object.

[0025] From the above description, it can be seen that when determining the operation object that matches the data management request, not only other hosts in the same distributed AI node as the client are considered, but also the hosts of other distributed AI nodes. In addition, the bandwidth usage, system status and file resource information contained in each IP resource are considered. The operation object that meets the data management request is comprehensively considered from multiple dimensions to ensure the effectiveness and reliability of data management.

[0026] Furthermore, the data management request includes one or more of a file download request, a file update request and a file deletion request.

[0027] It can be seen from the above description that the data management request includes one or more of a file download request, a file update request and a file deletion request, which ensures the diversity of data management and increases the scenarios to which data management can be applied.

[0028] Further, the data management request includes a file download request, and the file download request includes an identifier of a file to be downloaded;

[0029] The S32 includes:

[0030] Determine a matching host according to the identifier of the file to be downloaded, the bandwidth usage of each IP resource of the host added to the source distributed AI node contained in the source distributed AI node, the system status, the previous distributed AI node identifier and the next distributed AI node identifier of the source distributed AI node, and the file resource information contained in each IP resource in the IP resource pool of the involved distributed AI node;

[0031] The matched host IP is sent to the host where the source IP is located to perform a file download operation.

[0032] From the above description, it can be seen that when downloading a file, in order to match the target file, a comprehensive search is performed in the data source warehouse composed of multiple distributed AI nodes, and the bandwidth usage and system status of the IP resources where the target file is located are taken into consideration, thereby determining the most matching target file for download, thereby ensuring the efficiency of file downloading.

[0033] Further, the data management request includes a file update request, and the file update request includes an identifier of a file to be updated;

[0034] The S32 includes:

[0035] Determine a matching host according to the file identifier to be updated, the previous distributed AI node identifier and the next distributed AI node identifier contained in the source distributed AI node, and the file resource information contained under each IP resource in the IP resource pool of the distributed AI node involved;

[0036] The corresponding file on the matching host is updated according to the file update request.

[0037] From the above description, it can be seen that when updating a file, the previous distributed AI node identifier and the next distributed AI node identifier contained in the source distributed AI node to which the source IP belongs are used to perform a comprehensive search of the data source warehouse composed of multiple distributed AI nodes, thereby ensuring the reliability and accuracy of the file update.

[0038] Further, the data management request includes a file deletion request, and the file deletion request includes an identifier of a file to be deleted;

[0039] The S32 includes:

[0040] Determine a matching host according to the file identifier to be deleted, the previous distributed AI node identifier and the next distributed AI node identifier contained in the source distributed AI node, and the file resource information contained under each IP resource in the IP resource pool of the distributed AI node involved;

[0041] Delete the corresponding file on the matched host according to the file deletion request.

[0042] From the above description, it can be seen that when deleting files, the previous distributed AI node identifier and the next distributed AI node identifier contained in the source distributed AI node to which the source IP belongs are used to achieve a comprehensive retrieval of the data source warehouse composed of multiple distributed AI nodes, thereby ensuring the reliability and accuracy of file deletion.

[0043] Furthermore, after S3, the step of:

[0044] S4. Receive an operation update request sent by the client, and update data features of a distributed AI node where the host involved in the operation is located according to the operation update request.

[0045] From the above description, it can be seen that in the process of executing data management operations, the data characteristics of the distributed AI nodes where the host involved in the operation is located are updated according to the client's operation update request, so as to realize multi-dimensional management of files, and timely update the data characteristics of the distributed AI nodes according to the operations in the data management process, thereby ensuring the timeliness of the data management operation.

[0046] Furthermore, the process of performing the operation corresponding to the data management request according to the operation object also includes:

[0047] Receiving a connection request sent by the client to connect to a target host, wherein the connection request includes an IP address of the target host;

[0048] Determine, according to the source IP and the IP address of the target host, a relay distributed AI node that can be linked to both the source IP and the IP address of the target host;

[0049] The relay distributed AI node identifier is sent to the client.

[0050] From the above description, it can be seen that in the process of performing data management operations, when the client cannot connect to the target host, the relay distributed AI node that can connect to both the client and the target host can be used as a springboard to achieve the connection between the client and the target host, thereby further expanding the application scenarios of data management.

[0051] Furthermore, the data features include file name information, file structure information, data content information and data source information;

[0052] The operation update request includes an updated data type, and the updated data type includes one or more of the file name information, file structure information, data content information, and data source information;

[0053] The updating of the data features of the distributed AI node where the host involved in the operation is located according to the operation update request includes:

[0054] Update the data features of the corresponding type of the distributed AI node where the host involved in the operation is located according to the updated data type.

[0055] From the above description, it can be seen that data features include file name information, file structure information, data content information and data source information. It can not only perform data management in multiple dimensions, such as file name, data content fragment, file structure and data source information, but also realize operation updates in multiple dimensions, thereby improving the flexibility of operation updates. At the same time, after executing the data management operation, updates related to the data management operation are executed according to the user's wishes, which can further ensure the security of the data. The update is only executed when the user allows it.

[0056] like Figure 2 As shown, a distributed AI-based data management terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, each step in the above-mentioned distributed AI-based data management method is implemented.

[0057] The above-mentioned distributed AI-based data management method and terminal of the present application can be applied to various application scenarios that require data management, such as company-level file storage, query, download, update, deletion, etc., which are described below through specific implementation methods:

[0058] Embodiment 1

[0059] like Figure 1 As shown, a data management method based on distributed AI includes the following steps:

[0060] S1. Set up multiple distributed AI nodes, each distributed AI node includes an IP resource pool and data features, the IP resource pool includes IP resource information of hosts added to the distributed AI node, and the data features include file resource information included under each IP resource in the IP resource pool;

[0061] S2. Receive a data management request from a client, where the data management request includes a source IP address to which the client belongs;

[0062] S3. Determine the operation object that matches the data management request based on the source IP and each distributed AI node and its corresponding IP resource pool and data characteristics, and perform the operation corresponding to the data management request based on the operation object.

[0063] Among them, Figure 3 As shown, the IP resource information includes bandwidth usage, system status, and the previous distributed AI node identifier and next distributed AI node identifier of each IP resource under each IP resource; the data features include file name information, file structure information, data content information, and data source information;

[0064] The IP resource pool contains the following for each IP environment:

[0065] (1) System bandwidth: bandwidth usage, for example, whether it is high-traffic or low-traffic, and the maximum bandwidth value;

[0066] (2) System status: such as busy, starting to be busy, idle, offline, etc.

[0067] (3) It contains the previous distributed AI node and the next distributed AI node, that is, it is possible to query the upper and lower distributed AI node information through the current distributed AI node where the client is located. In specific implementation, the previous distributed AI node identifier and the next distributed AI node identifier of the current distributed AI node can be recorded in the IP resource pool to facilitate query;

[0068] (4) Request: refers to a data management request sent by a client, i.e., a user operation type. The data management request includes one or more of a file download request, a file update request, and a file deletion request;

[0069] Data features refer to the file resource information contained in each IP resource:

[0070] (1) File name information: a simple file name, which can be presented in MD5 format;

[0071] (2) File structure information: refers to the members of a file. For example, a file originally in the directory above is now moved to the current directory and can be presented in the form of a hash.

[0072] (3) Data content information: refers to the specific file content;

[0073] (4) Data source information: refers to the location of the data, such as which host or IP resource it is located on;

[0074] Figure 3 In the above, user refers to the user who participates in the data management of distributed AI. The user can log in to the host to achieve the connection, thereby joining the distributed data management composed of multiple distributed AI nodes. Once the user is connected to the host, the host will transmit the user's IP to the distributed AI node where the host is located for integration. When the user sends a data management request, the current distributed AI node will save the request of the currently joined user: file download, file deletion, file update or a combination of the above requests, and record the corresponding request in the "request" in the IP resource pool. At the user's request, the data content involved in this request can also be written into the data characteristics of the distributed AI node;

[0075] The S3 includes:

[0076] S31. Determine the source distributed AI node to which the source IP belongs according to the source IP;

[0077] S32, determining an operation object matching the data management request according to the data management request, bandwidth usage of each IP resource of the host added to the source distributed AI node contained in the source distributed AI node, system status, the previous distributed AI node identifier and the next distributed AI node identifier of the source distributed AI node, and file resource information contained in each IP resource in the IP resource pool of the involved distributed AI node, and performing an operation corresponding to the data management request according to the operation object;

[0078] In the specific implementation, first determine what request the client sends: file download, update, delete, or a combination of the above, and search for data that meets the request conditions in the data features. Then determine the system status of the host where the data is located. If it is in an idle state, obtain the system bandwidth of the host in the target IP resource pool, and search for the IP resource that meets the request conditions and matches the client best by traversing the upper and lower distributed AI nodes. After the match is completed, send the optimal IP resource to the client, so that the client can connect to the host corresponding to the optimal IP resource to complete the relevant data management request.

[0079] Embodiment 2

[0080] This embodiment describes how to download a file, specifically:

[0081] The data management request includes a file download request, and the file download request includes an identifier of a file to be downloaded;

[0082] The S32 includes:

[0083] Determine a matching host according to the identifier of the file to be downloaded, the bandwidth usage of each IP resource of the host added to the source distributed AI node contained in the source distributed AI node, the system status, the previous distributed AI node identifier and the next distributed AI node identifier of the source distributed AI node, and the file resource information contained in each IP resource in the IP resource pool of the involved distributed AI node;

[0084] Send the matched host IP to the host where the source IP is located to perform a file download operation;

[0085] For example: Xiaodong needs to download a folder consisting of 10 files (f0-f9). Since requests to a target host at the same time will inevitably affect the target bandwidth, first, when the distributed AI node where the source host where Xiaodong logs in is located receives a download request containing 10 files (f0-f9), it determines the total bandwidth of Xiaodong and matches the data features corresponding to the distributed AI nodes in turn. After the matching is completed and the conditions are met, the IP that needs to be linked is returned. For example: 10 files are matched to 2 distributed AI nodes, of which there are 20 hosts under the 2 distributed AI nodes that meet the conditions. Through the source host and the current direct optimal principle (the lowest ping value and the other party's bandwidth is large enough), 10 or 5 of them are selected for downloading operations.

[0086] Embodiment 3

[0087] This embodiment describes how to update a file, specifically:

[0088] The data management request includes a file update request, and the file update request includes an identifier of a file to be updated;

[0089] The S32 includes:

[0090] Determine a matching host according to the file identifier to be updated, the previous distributed AI node identifier and the next distributed AI node identifier contained in the source distributed AI node, and the file resource information contained under each IP resource in the IP resource pool of the distributed AI node involved;

[0091] The corresponding file on the matching host is updated according to the file update request.

[0092] For example: after finding the target file according to the file identifier to be updated, the target file is contained on two hosts under one distributed AI node. When the user initiates an update request operation, the distributed AI node will establish a link between the source host and the target host, and overwrite the corresponding target file of the target host through the source host.

[0093] Embodiment 4

[0094] This embodiment describes how to delete a file, specifically:

[0095] The data management request includes a file deletion request, and the file deletion request includes an identifier of a file to be deleted;

[0096] The S32 includes:

[0097] Determine a matching host according to the file identifier to be deleted, the previous distributed AI node identifier and the next distributed AI node identifier contained in the source distributed AI node, and the file resource information contained under each IP resource in the IP resource pool of the distributed AI node involved;

[0098] Delete the corresponding file on the matched host according to the file deletion request.

[0099] When performing file update or deletion operations, after returning the qualified host IP, all updates or deletions can be performed based on the qualified IP. In another implementation, partial updates or deletions can also be performed. The specific criteria for selecting updates or deletions can refer to the bandwidth usage and system status of the host IP.

[0100] In another optional embodiment, the step S3 further includes the following steps:

[0101] S4. Receive an operation update request sent by the client, and update data features of a distributed AI node where the host involved in the operation is located according to the operation update request.

[0102] The operation update request includes an updated data type, and the updated data type includes one or more of the file name information, file structure information, data content information, and data source information;

[0103] The updating of the data features of the distributed AI node where the host involved in the operation is located according to the operation update request includes:

[0104] Update the data features of the corresponding type of the distributed AI node where the host involved in the operation is located according to the updated data type.

[0105] For example, in the third embodiment above, after the file update is executed, if an operation update request sent by the client is received, the distributed AI node can update the data content information in the current data feature of the corresponding host after the update operation is completed, and the update operation does not change the file name, file structure and file source information, so these contents do not need to be changed;

[0106] In another optional implementation, in order to ensure the real-time nature of the data content stored in the distributed AI nodes, it can also be arranged that once the client completes the relevant data management operation, the corresponding distributed AI node will execute the update of each data involved in the data management operation at the same time, that is, when the source host (current user) and the target host complete downloading files, deleting files, updating files or combined operations, the source host's operation: downloading files, deleting files, updating files or combined operations will be notified to the corresponding distributed AI node, and the distributed AI node will record the user operation type involved in this operation: downloading files, deleting files, updating files or combined status and append the IP resource pool information and data feature information of the distributed AI node to which the host involved belongs.

[0107] Since data features include file name information, file structure information, data content information, and data source information, when searching for files, you can query in multiple dimensions, either based on the file name, the specific file content, the file storage location, or the host IP where the file is located. For example:

[0108] Employee A forgot to organize the document that her leader asked her to do. She didn't know the name or content, and she had to hand it in tomorrow. Suddenly, she remembered that the data features of the distributed AI nodes have a data source information function. She remembered the IP address of the computer on the day of the leader's meeting, so she found the IP address through the data source information and the content of the meeting on that day through the time and date, and successfully completed the document organization.

[0109] Embodiment 5

[0110] This embodiment describes how to connect two hosts that cannot be connected by using distributed AI nodes. Specifically:

[0111] The process of performing the operation corresponding to the data management request according to the operation object further includes:

[0112] Receiving a connection request sent by the client to connect to a target host, wherein the connection request includes an IP address of the target host;

[0113] Determine, according to the source IP and the IP address of the target host, a relay distributed AI node that can be linked to both the source IP and the IP address of the target host;

[0114] The relay distributed AI node identifier is sent to the client.

[0115] When a user cannot join a distributed AI node normally, the user can connect to it through other distributed AI node hosts:

[0116] For example: 111.168.105.1 is the target host IP;

[0117] Your IP: 101.108.100.100 cannot connect to 111.168.105.1;

[0118] Your colleague's IP: 131.208.150.150 can connect to 111.168.105.1. The distributed AI node uses the IP link between you and your colleague as a springboard, so that you can access 111.168.105.1 and complete related operations.

[0119] Distributed AI nodes can keep all hosts connected, such as:

[0120] Node N corresponds to hosts n1-n3;

[0121] Node F corresponds to f1-f3;

[0122] Node G corresponds to g1-g3;

[0123] N contains the information of the next node F, and F contains the information of G. n1 in N wants to access the information of g1. At this time, it is found that n1 cannot directly access the information of g1, but f1 can access the information of g1, and n1 and f1 can also access it. At this time, n1 links f1 and g1 with the help of F node, and completes the related operations after linking with n1.

[0124] The actual application scenarios are as follows:

[0125] Xiao Zhang is on a business trip in Beijing and needs to complete work-related content through a distributed AI node. Xiao Hong is on a business trip abroad and her work is very important and she needs to complete synchronization operations through distributed AI nodes in real time. However, the foreign network cannot update the target user host data. Xiao Hong is very anxious. When Xiao Hong knows that Xiao Zhang is on a business trip in Beijing, she tries to connect through the distributed AI node where Xiao Zhang is. As a result, the connection is established. Through the IP resource pool springboard to which Xiao Zhang's distributed AI node belongs, she successfully accesses the target user host and completes related file operations. At the same time, the distributed AI node records the data source information of Xiao Hong and Xiao Zhang's operations.

[0126] Embodiment 6

[0127] Please refer to Figure 2 , a distributed AI-based data management terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, each step of a distributed AI-based data management method described in any one of Embodiments 1 to 5 above is implemented.

[0128] In summary, the present invention provides a distributed AI-based data management method and terminal, which sets multiple distributed AI nodes, each distributed AI node includes IP resource information of each host joining the node and file resource information contained under each IP resource. When a data management request from a client is received, the source IP to which the client belongs is determined, and an operation object matching the data management request is determined based on the source IP and each distributed AI node and its corresponding IP resource pool and data characteristics. The operation corresponding to the data management request is performed according to the operation object. Only clients joining the distributed AI nodes can perform data management within the range of hosts included in each distributed AI node, and when performing data management, the comprehensive The IP resource information of each host and the file resource information contained in each IP resource are considered together. The IP resource information includes the bandwidth usage, system status, and the previous distributed AI node identifier and the next distributed AI node identifier of each IP resource. The file resource information includes file name information, file structure information, data content information, and data source information. With the help of the intelligent nature of AI, AI technology is used based on distributed AI nodes to achieve safe and reasonable allocation of resources. A distributed data source warehouse is formed by distributed AI nodes, which can realize multi-dimensional file management, ensure the security and reliability of file data management, and can be applied to multiple scenarios, such as company-level file data management.

[0129] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A data management method based on distributed AI, characterized in that: Includes steps: S1. Set up multiple distributed AI nodes, each distributed AI node includes an IP resource pool and data features, the IP resource pool includes IP resource information of hosts added to the distributed AI node, and the data features include file resource information included under each IP resource in the IP resource pool; S2. Receive a data management request from a client, where the data management request includes a source IP address to which the client belongs; S3. Determine the operation object that matches the data management request based on the source IP and each distributed AI node and its corresponding IP resource pool and data characteristics, and perform the operation corresponding to the data management request based on the operation object.

2. A distributed AI-based data management method according to claim 1, characterized in that: The IP resource information includes bandwidth usage, system status, and the previous distributed AI node identifier and the next distributed AI node identifier of each IP resource; The S3 includes: S31. Determine the source distributed AI node to which the source IP belongs according to the source IP; S32. Determine the operation object that matches the data management request based on the data management request, the bandwidth usage of each IP resource of the host that joins the source distributed AI node, the system status, the previous distributed AI node identifier and the next distributed AI node identifier of the source distributed AI node, and the file resource information contained in each IP resource in the IP resource pool of the distributed AI node involved, and perform the operation corresponding to the data management request based on the operation object.

3. A distributed AI-based data management method according to claim 2, characterized in that: The data management request includes one or more of a file download request, a file update request and a file deletion request.

4. A distributed AI-based data management method according to claim 3, characterized in that: The data management request includes a file download request, and the file download request includes an identifier of a file to be downloaded; The S32 includes: Determine a matching host according to the identifier of the file to be downloaded, the bandwidth usage of each IP resource of the host added to the source distributed AI node contained in the source distributed AI node, the system status, the previous distributed AI node identifier and the next distributed AI node identifier of the source distributed AI node, and the file resource information contained in each IP resource in the IP resource pool of the involved distributed AI node; The matched host IP is sent to the host where the source IP is located to perform a file download operation.

5. The distributed AI-based data management method according to claim 3, characterized in that: The data management request includes a file update request, and the file update request includes an identifier of a file to be updated; The S32 includes: Determine a matching host according to the file identifier to be updated, the previous distributed AI node identifier and the next distributed AI node identifier contained in the source distributed AI node, and the file resource information contained under each IP resource in the IP resource pool of the distributed AI node involved; The corresponding file on the matching host is updated according to the file update request.

6. The data management method based on distributed AI according to claim 3, characterized in that: The data management request includes a file deletion request, and the file deletion request includes an identifier of a file to be deleted; The S32 includes: Determine a matching host according to the file identifier to be deleted, the previous distributed AI node identifier and the next distributed AI node identifier contained in the source distributed AI node, and the file resource information contained under each IP resource in the IP resource pool of the distributed AI node involved; Delete the corresponding file on the matched host according to the file deletion request.

7. A distributed AI-based data management method according to any one of claims 4 to 6, characterized in that: The S3 further comprises the following steps: S4. Receive an operation update request sent by the client, and update data features of a distributed AI node where the host involved in the operation is located according to the operation update request.

8. A distributed AI-based data management method according to any one of claims 4 to 6, characterized in that: The process of performing the operation corresponding to the data management request according to the operation object further includes: Receiving a connection request sent by the client to connect to a target host, wherein the connection request includes an IP address of the target host; Determine, according to the source IP and the IP address of the target host, a relay distributed AI node that can be linked to both the source IP and the IP address of the target host; The relay distributed AI node identifier is sent to the client.

9. The distributed AI-based data management method according to claim 7, characterized in that: The data features include file name information, file structure information, data content information and data source information; The operation update request includes an updated data type, and the updated data type includes one or more of the file name information, file structure information, data content information, and data source information; The updating of the data features of the distributed AI node where the host involved in the operation is located according to the operation update request includes: Update the data features of the corresponding type of the distributed AI node where the host involved in the operation is located according to the updated data type.

10. A distributed AI-based data management terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, each step in the distributed AI-based data management method as described in any one of claims 1 to 9 is implemented.

Images