Task allocation method and device for a system integrating synaesthesia, computing and storage

Abstract

The present invention provides a task allocation method and device for a system integrating synaesthesia, computing and storage, and belongs to the field of resource allocation. The system integrating synaesthesia, computing and storage includes a control node and a plurality of computing nodes, and the method is applied to the control node; the method includes: the system integrating synaesthesia, computing and storage includes a control node and a plurality of computing nodes, and the method is applied to the control node; the method includes: obtaining a task to be allocated; wherein the task to be allocated is composed of a plurality of subtasks; extracting task features from each subtask to obtain a task feature label of each subtask; calculating the matching degree of the task feature label of each subtask with the historical feature label of each computing node, and using it as the matching degree of each subtask with each computing node; determining the target computing node corresponding to each subtask based on the matching degree of each subtask with each computing node, so as to allocate each subtask. The present invention can ensure the efficiency and effect of task processing.

Description

Technical Field

[0001] The present invention relates to the technical field of resource scheduling, and in particular to a task allocation method and device for a system integrating sensing, computing and storage. Background Art

[0002] The integrated system of synergy, computing and storage is composed of multiple nodes such as base stations and terminals, and has communication, perception, computing and storage functions. The integrated system of synergy, computing and storage combines hardware with different process architectures, different instruction sets and different functions to improve the computing power level, and interconnects and realizes heterogeneous computing.

[0003] However, currently, computing tasks are usually allocated by only considering the communication, computing power, energy consumption and other characteristics of each node, without considering the adaptability of different hardware to process different types of tasks. If tasks are assigned to nodes that do not have processing capabilities, the task processing time will be extended and the processing effect will be poor. Summary of the invention

[0004] The embodiment of the present invention provides a task allocation method and device for a system integrating sensing, computing and storage, so as to solve the problem of unreasonable node task allocation.

[0005] In a first aspect, an embodiment of the present invention provides a task allocation method for a system integrating synergy, computing and storage, wherein the system integrating synergy, computing and storage includes a control node and a plurality of computing nodes, and the method is applied to the control node; the method includes:

[0006] Obtaining tasks to be assigned; wherein the tasks to be assigned are composed of multiple subtasks;

[0007] Extract task features for each subtask to obtain task feature labels for each subtask;

[0008] Calculate the matching degree between the task feature label of each subtask and the historical feature label of each computing node, and use it as the matching degree between each subtask and each computing node; wherein the historical feature label is determined based on the task feature label of the historical task of the corresponding computing node;

[0009] The target computing node corresponding to each subtask is determined based on the matching degree between each subtask and each computing node, so as to allocate each subtask.

[0010] In a possible implementation, task features are extracted for each subtask, including:

[0011] For each subtask, the model feature vector of the subtask is extracted, and based on the vector similarity algorithm and the model feature vector of the subtask, a preset number of related model feature vectors are determined in the model feature vector set corresponding to the task type of the subtask, and a task feature label of the subtask is generated based on each related model feature vector; wherein the task types include data compression, data decompression, image processing, numerical calculation, data encryption and data decryption, and the model feature set corresponding to each task type is constructed based on the model feature vectors of multiple historical tasks belonging to the task type, and each model feature vector corresponds to a task feature label.

[0012] In a possible implementation, extracting the model feature vector of the subtask includes:

[0013] Structuring the text content of the subtask to obtain a structured text of the subtask;

[0014] The structured text of the subtask is vectorized to obtain the model feature vector of the subtask.

[0015] In a possible implementation, calculating the matching degree between the task feature label of each subtask and the historical feature label of each candidate computing node includes:

[0016] For each subtask, the computing node whose historical feature label contains all the task feature labels of the subtask is taken as the candidate computing node of the subtask, and the matching degree between the subtask and the candidate computing node is calculated based on the proficiency and efficiency scores of each task feature label in each candidate computing node of the subtask.

[0017] In a possible implementation, before calculating the matching degree between the task feature label of each subtask and the historical feature label of each computing node and using the matching degree between each subtask and each computing node as the matching degree, the method further includes:

[0018] For each computing node, the historical feature labels of the candidate computing nodes are determined based on the task feature labels of the historical tasks of the computing node, the proficiency of each task feature label in the computing node is determined based on the number of occurrences and the task completion time, and the efficiency score of the task feature label in the computing node is determined based on the task processing time and task volume of each task feature label.

[0019] In a possible implementation, determining a target computing node corresponding to each subtask based on the matching degree between each subtask and each computing node, so as to allocate each subtask, includes:

[0020] The subtasks are sorted based on the number of task feature labels and the task volume of each subtask, and the target computing node corresponding to each subtask is determined in sequence.

[0021] In a possible implementation, the subtasks are sorted based on the number of task feature tags and the amount of tasks of each subtask, including:

[0022] Sort each subtask in descending order according to the product of the number of task feature labels and the task amount.

[0023] In a possible implementation, determining the target computing node corresponding to each subtask in sequence includes:

[0024] For each subtask, the target computing node corresponding to the subtask is determined based on the matching degree between the subtask and each computing node and the amount of tasks already allocated to each computing node.

[0025] In a second aspect, an embodiment of the present invention provides a task allocation device for a system integrating synergy, computing and storage, wherein the system integrating synergy, computing and storage comprises a control node and a plurality of computing nodes, and the device is applied to the control node; the device comprises:

[0026] An acquisition module is used to acquire tasks to be assigned; wherein the tasks to be assigned are composed of multiple subtasks;

[0027] The extraction module is used to extract the task features of each subtask and obtain the task feature labels of each subtask;

[0028] A matching module, used to calculate the matching degree between the task feature label of each subtask and the historical feature label of each computing node, and use it as the matching degree between each subtask and each computing node; wherein the historical feature label is determined based on the task feature label of the historical task of the corresponding computing node;

[0029] The allocation module is used to determine the target computing node corresponding to each subtask based on the matching degree between each subtask and each computing node, so as to allocate each subtask.

[0030] The embodiment of the present invention provides a task allocation method and device for a system integrating synaesthesia, computing and storage, which extracts task features of subtasks in the assigned tasks to determine task feature labels to indicate the computing model or computing capability that a node needs to possess to process each subtask, and determines the historical feature labels of each computing node through the historical tasks of each computing node in the system to indicate the computing model or computing capability that the computing node actually possesses. The target computing node of each subtask is determined through the task feature labels of the subtasks and the historical feature labels of each computing node, so that the subtasks can be reasonably allocated to ensure the efficiency and effect of task processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0032] Figure 1 It is a flow chart for implementing a task allocation method of a synaesthesia computing and storage integrated system provided by an embodiment of the present invention;

[0033] Figure 2 It is a structural schematic diagram of a task allocation device of a synaesthesia-computing-storage-integrated system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0034] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0035] In order to make the purpose, technical solutions and advantages of the present invention more clear, specific embodiments will be described below in conjunction with the accompanying drawings.

[0036] Figure 1 The implementation flow chart of the task allocation method of the integrated system of synergy computing and storage provided by the embodiment of the present invention, the integrated system of synergy computing and storage includes a control node and multiple computing nodes, and the method is applied to the control node, which is described in detail as follows:

[0037] Step 101, obtaining a task to be assigned; wherein the task to be assigned is composed of a plurality of subtasks.

[0038] In this embodiment, the tasks to be assigned can be initiated by one or more computing nodes in the integrated computing and storage system and assigned by the control node. Each subtask has a task type, which facilitates the control node to determine the computing node suitable for assignment according to the task type.

[0039] Step 102: extract task features for each subtask to obtain task feature labels for each subtask.

[0040] In this embodiment, the task feature label of each subtask can represent the model and capabilities required to process the subtask. The hardware structure, instruction architecture and built-in algorithm model of each computing node are different. The task feature label can be used to match the computing node suitable for processing each subtask, thereby making a reasonable allocation.

[0041] Step 103, calculate the matching degree between the task feature label of each subtask and the historical feature label of each computing node, and use it as the matching degree between each subtask and each computing node; wherein the historical feature label is determined based on the task feature label of the historical task of the corresponding computing node.

[0042] In this embodiment, the historical tasks of the computing node refer to the tasks that the computing node has completed and are also the tasks that the computing node is capable of processing. Therefore, the task feature labels of the historical tasks of the computing node can represent the task processing capabilities of the computing node and are used to match the task feature labels of the subtasks.

[0043] Step 104 : determining the target computing node corresponding to each subtask based on the matching degree between each subtask and each computing node, so as to allocate each subtask.

[0044] In this embodiment, the higher the matching degree between the subtask and the computing node, the stronger the ability of the computing node to process the subtask. A suitable target computing node can be selected based on the matching degree.

[0045] To ensure balanced distribution of tasks, the existing task volume of each computing node can also be considered during the distribution.

[0046] The embodiment of the present invention extracts task features of subtasks in the assigned task to determine task feature labels to indicate the computing model or computing capability that the node needs to possess in order to process each subtask, and determines the historical feature labels of each computing node through the historical tasks of each computing node in the system to indicate the computing model or computing capability that the computing node actually possesses. The target computing node of each subtask is determined through the task feature labels of the subtasks and the historical feature labels of each computing node, so that the subtasks can be reasonably allocated to ensure the efficiency and effectiveness of task processing.

[0047] In a possible implementation, task features are extracted for each subtask, including:

[0048] For each subtask, the model feature vector of the subtask is extracted, and based on the vector similarity algorithm and the model feature vector of the subtask, a preset number of related model feature vectors are determined in the model feature vector set corresponding to the task type of the subtask, and a task feature label of the subtask is generated based on each related model feature vector; wherein the task types include data compression, data decompression, image processing, numerical calculation, data encryption and data decryption, and the model feature set corresponding to each task type is constructed based on the model feature vectors of multiple historical tasks belonging to the task type, and each model feature vector corresponds to a task feature label.

[0049] In this embodiment, the Faiss algorithm can be used to determine the relevant model feature vector. The historical tasks corresponding to the relevant model feature vector have a high similarity with the parameter logical relationship of the subtask. It can be considered that the model or processing capability required for these historical tasks is the same as that of the subtask, and the task feature label corresponding to the historical figure can be used as the task feature label of the subtask.

[0050] In a possible implementation, extracting the model feature vector of the subtask includes:

[0051] Structuring the text content of the subtask to obtain a structured text of the subtask;

[0052] The structured text of the subtask is vectorized to obtain the model feature vector of the subtask.

[0053] In this embodiment, after structural processing, the logical operation relationship between the data and parameters in the subtask can be extracted, and then the text can be vectorized to facilitate semantic processing and determine the model suitable for the subtask.

[0054] In a possible implementation, calculating the matching degree between the task feature label of each subtask and the historical feature label of each candidate computing node includes:

[0055] For each subtask, the computing node whose historical feature label contains all the task feature labels of the subtask is taken as the candidate computing node of the subtask, and the matching degree between the subtask and the candidate computing node is calculated based on the proficiency and efficiency scores of each task feature label in each candidate computing node of the subtask.

[0056] In this embodiment, a higher proficiency means that the candidate computing node has processed more subtasks of the same type, and thus has a stronger ability to process the subtask. A higher efficiency score means that the candidate computing node has a higher processing efficiency for the subtask.

[0057] The matching degree between the subtask and the candidate computing node = the sum of the proficiency of each task feature label * proficiency weight coefficient + efficiency score * efficiency weight coefficient. Each weight coefficient can be set according to the actual situation.

[0058] In a possible implementation, before calculating the matching degree between the task feature label of each subtask and the historical feature label of each computing node and using the matching degree between each subtask and each computing node as the matching degree, the method further includes:

[0059] For each computing node, the historical feature labels of the candidate computing nodes are determined based on the task feature labels of the historical tasks of the computing node, the proficiency of each task feature label in the computing node is determined based on the number of occurrences and the task completion time, and the efficiency score of the task feature label in the computing node is determined based on the task processing time and task volume of each task feature label.

[0060] In this embodiment, the number of occurrences of the task feature label indicates the number of times the computing node completes the task with the number of occurrences of the task feature label. The more times the task is completed, the higher the proficiency. The later the task completion time is, the later the computing node completes the assigned task. The model or algorithm configured in the computing node belongs to a newer version with more complete functions. The task completion time can be set to correspond to different coefficients in different time periods. The larger the time period, the larger the coefficient.

[0061] The proficiency of the task feature label = the number of occurrences of the task feature label + the coefficient corresponding to the task completion time.

[0062] The efficiency score of a task feature label = task volume / task processing time. The higher the efficiency score of a task feature label, the higher the processing efficiency of the computing node for tasks with such task feature labels.

[0063] In a possible implementation, determining a target computing node corresponding to each subtask based on the matching degree between each subtask and each computing node, so as to allocate each subtask, includes:

[0064] The subtasks are sorted based on the number of task feature labels and the task volume of each subtask, and the target computing node corresponding to each subtask is determined in sequence.

[0065] In this embodiment, after a subtask is assigned to a computing node, the computing resources of the computing node are occupied. It is necessary to avoid assigning subsequent subtasks to the same computing node as much as possible to avoid excessive concentration of computing tasks on the same computing node, which leads to reduced efficiency. Therefore, the subtasks can be sorted and assigned in order. After each subtask is assigned, the assigned task information of the computing node is updated as reference information for subsequent assignments.

[0066] In a possible implementation, the subtasks are sorted based on the number of task feature tags and the amount of tasks of each subtask, including:

[0067] Sort each subtask in descending order according to the product of the number of task feature labels and the task amount.

[0068] In this embodiment, the larger the product of the number of task feature tags and the task amount, the more complex the subtask is, the higher the processing capability and efficiency requirements for the computing node are, and the subtask should be allocated first.

[0069] In a possible implementation, determining the target computing node corresponding to each subtask in sequence includes:

[0070] For each subtask, the target computing node corresponding to the subtask is determined based on the matching degree between the subtask and each computing node and the amount of tasks already allocated to each computing node.

[0071] In this embodiment, the computing nodes to which the amount of assigned tasks is greater than the threshold can be removed from each computing node, and the target computing node corresponding to the subtask can be determined from the remaining computing nodes based on the matching degree. For example, after subtask 1 is assigned to computing node A, when determining the target computing node of subtask 2, computing node A is removed from the candidate computing nodes; then, after subtask 22 is assigned to computing node B, the target computing node of subtask 3 needs to be determined, and computing node B is removed from the candidate computing nodes and assigned, so as to avoid continuously assigning multiple subtasks to the same computing node.

[0072] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0073] The following is an embodiment of the device of the present invention. For details not described in detail therein, reference may be made to the corresponding method embodiment described above.

[0074] Figure 2 The structure diagram of the task allocation device of the system of integrated synaesthesia, computing and storage provided by an embodiment of the present invention is shown. For the convenience of explanation, only the part related to the embodiment of the present invention is shown, which is described in detail as follows:

[0075] like Figure 2 As shown, the task allocation device 2 of the system integrating synaesthesia, computing and storage includes:

[0076] The acquisition module 21 is used to acquire the task to be assigned; wherein the task to be assigned is composed of multiple subtasks;

[0077] An extraction module 22 is used to extract task features of each subtask to obtain a task feature label of each subtask;

[0078] A matching module 23 is used to calculate the matching degree between the task feature label of each subtask and the historical feature label of each computing node, and use it as the matching degree between each subtask and each computing node; wherein the historical feature label is determined based on the task feature label of the historical task of the corresponding computing node;

[0079] The allocation module 24 is used to determine the target computing node corresponding to each subtask based on the matching degree between each subtask and each computing node, so as to allocate each subtask.

[0080] In a possible implementation, the extraction module 22 is specifically used for:

[0081] For each subtask, the model feature vector of the subtask is extracted, and based on the vector similarity algorithm and the model feature vector of the subtask, a preset number of related model feature vectors are determined in the model feature vector set corresponding to the task type of the subtask, and a task feature label of the subtask is generated based on each related model feature vector; wherein the task types include data compression, data decompression, image processing, numerical calculation, data encryption and data decryption, and the model feature set corresponding to each task type is constructed based on the model feature vectors of multiple historical tasks belonging to the task type, and each model feature vector corresponds to a task feature label.

[0082] In a possible implementation, the extraction module 22 is specifically used for:

[0083] Structuring the text content of the subtask to obtain a structured text of the subtask;

[0084] The structured text of the subtask is vectorized to obtain the model feature vector of the subtask.

[0085] In a possible implementation, the matching module 23 is specifically configured to:

[0086] For each subtask, the computing node whose historical feature label contains all the task feature labels of the subtask is taken as the candidate computing node of the subtask, and the matching degree between the subtask and the candidate computing node is calculated based on the proficiency and efficiency scores of each task feature label in each candidate computing node of the subtask.

[0087] In a possible implementation, the extraction module 22 is further configured to:

[0088] Before calculating the matching degree between the task feature labels of each subtask and the historical feature labels of each computing node and using it as the matching degree between each subtask and each computing node, for each computing node, the historical feature labels of the candidate computing nodes are determined based on the task feature labels of the historical tasks of the computing node, the proficiency of each task feature label in the computing node is determined based on the number of occurrences and the task completion time of the task, and the efficiency score of the task feature label in the computing node is determined based on the task processing time and task volume of each task feature label.

[0089] In a possible implementation, the allocation module 24 is specifically configured to:

[0090] The subtasks are sorted based on the number of task feature labels and the task volume of each subtask, and the target computing node corresponding to each subtask is determined in sequence.

[0091] In a possible implementation, the allocation module 24 is specifically configured to:

[0092] Sort each subtask in descending order according to the product of the number of task feature labels and the task amount.

[0093] In a possible implementation, the allocation module 24 is specifically configured to:

[0094] For each subtask, the target computing node corresponding to the subtask is determined based on the matching degree between the subtask and each computing node and the amount of tasks already allocated to each computing node.

[0095] The embodiment of the present invention extracts task features of subtasks in the assigned task to determine task feature labels to indicate the computing model or computing capability that the node needs to possess in order to process each subtask, and determines the historical feature labels of each computing node through the historical tasks of each computing node in the system to indicate the computing model or computing capability that the computing node actually possesses. The target computing node of each subtask is determined through the task feature labels of the subtasks and the historical feature labels of each computing node, so that the subtasks can be reasonably allocated to ensure the efficiency and effectiveness of task processing.

[0096] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0097] Those of ordinary skill in the art will appreciate that the templates, units, and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0098] If the module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the task allocation method embodiment of each of the above-mentioned systems of synaesthesia, calculation and storage can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory, random access memory, electric carrier signal, telecommunication signal and software distribution medium, etc.

[0099] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A task allocation method for a system integrating synaesthesia, computing and storage, characterized in that: The system of integrated sensing, computing and storage includes a control node and a plurality of computing nodes, and the method is applied to the control node; the method includes: Obtaining a task to be assigned; wherein the task to be assigned is composed of multiple subtasks; Extract task features for each subtask to obtain task feature labels for each subtask; Calculate the matching degree of the task feature label of each subtask and the historical feature label of each computing node, and use it as the matching degree of each subtask and each computing node; wherein the historical feature label is determined based on the task feature label of the historical task of the corresponding computing node; Determine the target computing node corresponding to each subtask based on the matching degree between each subtask and each computing node, so as to allocate each subtask; The extracting of task features for each subtask includes: For each subtask, extract the model feature vector of the subtask, determine a preset number of related model feature vectors in the model feature vector set corresponding to the task type of the subtask based on a vector similarity algorithm and the model feature vector of the subtask, and generate a task feature label for the subtask based on each related model feature vector; wherein the task types include data compression, data decompression, image processing, numerical calculation, data encryption and data decryption, and the model feature set corresponding to each task type is constructed based on the model feature vectors of multiple historical tasks belonging to the task type, and each model feature vector corresponds to a task feature label.

2. The task allocation method of the system of synaesthesia, computing and storage according to claim 1 is characterized in that: The model feature vector extracted for this subtask includes: Structuring the text content of the subtask to obtain a structured text of the subtask; The structured text of the subtask is vectorized to obtain the model feature vector of the subtask.

3. The task allocation method of the synaesthesia computing and storage system according to claim 1 is characterized in that: The calculating the matching degree between the task feature labels of each subtask and the historical feature labels of each candidate computing node includes: For each subtask, the computing node whose historical feature label contains all the task feature labels of the subtask is taken as the candidate computing node of the subtask, and the matching degree between the subtask and the candidate computing node is calculated based on the proficiency and efficiency scores of each task feature label in each candidate computing node of the subtask.

4. The task allocation method of the synaesthesia computing and storage system according to claim 3 is characterized in that: Before calculating the matching degree between the task feature label of each subtask and the historical feature label of each computing node and using the matching degree between each subtask and each computing node as the matching degree, the method further includes: For each computing node, the historical feature labels of the candidate computing nodes are determined based on the task feature labels of the historical tasks of the computing node, the proficiency of each task feature label in the computing node is determined based on the number of occurrences and the task completion time, and the efficiency score of the task feature label in the computing node is determined based on the task processing time and task volume of each task feature label.

5. The task allocation method of the synaesthesia computing and storage system according to claim 1 is characterized in that: The determining of the target computing node corresponding to each subtask based on the matching degree between each subtask and each computing node, so as to allocate each subtask, includes: The subtasks are sorted based on the number of task feature labels and the task volume of each subtask, and the target computing node corresponding to each subtask is determined in sequence.

6. The task allocation method of the synaesthesia computing and storage system according to claim 5 is characterized in that: The subtasks are sorted based on the number of task feature labels and the amount of tasks, including: Sort each subtask in descending order according to the product of the number of task feature labels and the task amount.

7. The task allocation method of the synaesthesia computing and storage system according to claim 5 is characterized in that: The step of sequentially determining the target computing node corresponding to each subtask includes: For each subtask, the target computing node corresponding to the subtask is determined based on the matching degree between the subtask and each computing node and the amount of tasks already allocated to each computing node.

8. A task allocation device for a system integrating synaesthesia, computing and storage, characterized in that: The system of integrated sensing, computing and storage includes a control node and a plurality of computing nodes, and the device is applied to the control node; the device includes: An acquisition module, used for acquiring a task to be assigned; wherein the task to be assigned is composed of a plurality of subtasks; The extraction module is used to extract the task features of each subtask and obtain the task feature labels of each subtask; A matching module, used to calculate the matching degree between the task feature label of each subtask and the historical feature label of each computing node, and use it as the matching degree between each subtask and each computing node; wherein the historical feature label is determined based on the task feature label of the historical task of the corresponding computing node; An allocation module, used to determine the target computing node corresponding to each subtask based on the matching degree between each subtask and each computing node, so as to allocate each subtask; The extraction module is specifically used for: For each subtask, extract the model feature vector of the subtask, determine a preset number of related model feature vectors in the model feature vector set corresponding to the task type of the subtask based on a vector similarity algorithm and the model feature vector of the subtask, and generate a task feature label for the subtask based on each related model feature vector; wherein the task types include data compression, data decompression, image processing, numerical calculation, data encryption and data decryption, and the model feature set corresponding to each task type is constructed based on the model feature vectors of multiple historical tasks belonging to the task type, and each model feature vector corresponds to a task feature label.

Images