Task processing method of computing power network and related device

Abstract

This application discloses a task processing method for a computing power network, relating to the field of computer technology. The method includes: acquiring computing power information of multiple nodes in the computing power network; making task distribution decisions based on the computing power information of multiple nodes, a linear programming algorithm, and multiple constraints to obtain the computing power nodes that actually execute each task; wherein the variables of the linear programming algorithm include the computation time rate and the number of service replicas of each computing power node; and sending each task to the corresponding computing power node so that each computing power node can process the received task. This method can improve the efficiency of task processing in a computing power network. This application also discloses a task processing device, a computing device, and a computer-readable storage medium for a computing power network, which have the above-mentioned beneficial effects.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a task processing method, task processing device, computing equipment, and computer-readable storage medium for a computing power network. Background Technology

[0002] In related technologies, computing power networks combine cloud and edge computing resources through communication technology to build a powerful computing resource pool, thereby providing more diversified services and ensuring sufficient computing and storage resources. Therefore, computing power networks have a higher probability of meeting the computing power requirements of computationally intensive and latency-sensitive tasks than cloud computing and edge computing, and can improve the average utilization rate of computing resources and the overall response time of tasks.

[0003] In recent years, the fields of cloud computing and edge computing have carried out research on computing power collaboration and achieved some successes. However, these achievements have made strong assumptions about the application scenarios when designing collaborative models and algorithms, mainly including: 1. Cloud / edge servers only have a single type of computing chip, mainly CPU; 2. Data transmission latency between edges is negligible, mainly considering the transmission latency between cloud and edge; 3. The number of computations required for each task is known, and the computation time of the task on the server is obtained by dividing the number of computations required by the task by the CPU clock speed; 4. The pattern of task arrival at the cloud-edge system follows a known probability distribution.

[0004] However, these assumptions lead to significant shortcomings when applying existing methods to computing networks, making it difficult to fully unleash the scalability of computing networks. These shortcomings are mainly manifested in the following ways: 1. Computing devices in computing networks are typically configured with multiple computing units, and with the development and promotion of artificial intelligence technology, the configuration scheme is mainly a combination of CPUs and GPUs; 2. To improve the efficiency of computing power collaboration and make the model closer to the real computing network environment, the data transmission latency between computing devices cannot be ignored and is related to network bandwidth, transmission distance, and packet size; 3. The number of computations required to complete a computing network task is difficult to estimate because, when the application code required for the task is not publicly available, the number of computations cannot be predicted in advance; even when the code is known, it may contain a large number of loop operations and conditional statements, and the time to exit the loop and the result of the conditional statement are unpredictable, making the number of computations unpredictable; 4. Computing networks serve multiple users, and the arrival pattern of tasks is almost unpredictable. Each customer has its own unique task generation pattern. While it may be possible to model the task production pattern of a single customer, the composite operation pattern of multiple customers remains difficult to predict.

[0005] Therefore, how to design an efficient and universally applicable task processing method and device for computing power networks, overcome the shortcomings of existing methods, improve the working efficiency and resource utilization of computing power networks, and improve the overall response time of tasks are key issues of concern to those in this field. Summary of the Invention

[0006] The purpose of this application is to provide a task processing method, task processing device, computing equipment, and computer-readable storage medium for a computing power network, thereby improving the efficiency of the computing power network in processing tasks.

[0007] A task processing method for a computing power network, comprising:

[0008] Obtain computing power information for all computing power nodes in the computing power network;

[0009] Based on the computing power information, linear programming algorithm, and multiple constraints, a task distribution decision is made for the received tasks to obtain the computing power nodes that actually execute each task; wherein, the variables of the linear programming algorithm include the computing time rate and the number of service replicas of each computing power node.

[0010] Each task is sent to its corresponding computing node so that the computing node can process the received task.

[0011] Preferably, the computing power information includes computing power usage, network transmission status, and the resource requirements of received tasks.

[0012] Preferably, the planning objective of the linear programming algorithm is to complete all tasks within the task set in the shortest possible time.

[0013] Preferably, sending each of the tasks to the corresponding computing power node includes:

[0014] The corresponding information of each task and the computing power nodes that actually execute each task is encapsulated into a decision result data packet;

[0015] The decision result data packet is broadcast in the computing power network.

[0016] Preferably, it further includes:

[0017] Other computing nodes determine the corresponding tasks based on the obtained decision result data packets.

[0018] Preferably, it further includes:

[0019] Receive tasks sent by other computing power nodes and add the received tasks to the task queue;

[0020] The corresponding tasks are executed sequentially based on the task queue to obtain the processing results.

[0021] Preferably, based on the computing power information, linear programming algorithm, and multiple constraints, a task distribution decision is made for the received tasks to obtain the computing power nodes that actually execute each task, including:

[0022] Obtain the computing time rate and service replica count for each computing node;

[0023] Based on the computing power information, the linear programming algorithm, and the multiple constraints, the problem of computing power collaborative scheduling and processing tasks is modeled as an integer linear programming problem;

[0024] Solve the integer linear programming problem to obtain the computing nodes that actually execute each task.

[0025] A task processing device for a computing network, comprising:

[0026] The computing power information acquisition module is used to acquire computing power information of all computing power nodes in the computing power network;

[0027] The task planning module is used to make task distribution decisions on the received tasks based on the computing power information, linear programming algorithm and multiple constraints, so as to obtain the computing power nodes that actually execute each task; wherein, the variables of the linear programming algorithm include the computing time rate and the number of service replicas of each computing power node.

[0028] The task forwarding module is used to send each of the tasks to the corresponding computing power nodes so that each computing power node can process the received tasks.

[0029] This application also provides a computing device, including:

[0030] Memory, used to store computer programs;

[0031] A processor is configured to implement the task processing method described above when executing the computer program.

[0032] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the task processing method described above.

[0033] This application provides a task processing method for a computing power network, comprising: acquiring computing power information of all computing power nodes in the computing power network; making task distribution decisions on received tasks based on the computing power information, a linear programming algorithm, and multiple constraints to obtain the computing power nodes that actually execute each task; wherein, the variables of the linear programming algorithm include the computing time rate and the number of service replicas of each computing power node; and sending each task to the corresponding computing power node so that each computing power node can process the received task.

[0034] By acquiring computing power information from multiple computing nodes, and then making task distribution decisions based on this information, a linear programming algorithm, and multiple constraints, the corresponding computing node for each task is determined. Since the parameters of this linear programming algorithm include the computation time rate and the number of service replicas, which can evaluate the performance of computing nodes in the computing network for a specific application, when designing the optimal solution for collaborative task processing, only the runtime performance of the computing nodes needs to be considered, without taking into account the heterogeneous hardware configuration and software environment of the computing nodes in the computing network. This improves the efficiency of the computing network in processing tasks and increases the utilization rate of computing resources.

[0035] This application also provides a task processing device, computing device, and computer-readable storage medium for a computing power network, which have the above-mentioned beneficial effects, and will not be elaborated here. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0037] Figure 1 A flowchart illustrating a task processing method for a computing network provided in this application embodiment;

[0038] Figure 2 A detailed flowchart of a task processing method for a computing power network provided in an embodiment of this application;

[0039] Figure 3 This is a schematic diagram of the structure of a task processing device for a computing network provided in an embodiment of this application;

[0040] Figure 4 This application provides a schematic diagram of the structure of a computing device. Detailed Implementation

[0041] The core of this application is to provide a task processing method, task processing device, computing equipment, and computer-readable storage medium for computing power networks, thereby improving the efficiency of task processing in computing power networks.

[0042] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0043] This application provides a task processing method for a computing power network. First, each computing node in the network, acting as a source node, receives tasks submitted by users and then sends its own computing power information and the received task information to the computing power network management node. The management node, after obtaining the computing power information of each computing node and the task information collected by them, inputs this information into a pre-designed linear programming model. It then uses an existing solver to calculate the task distribution strategy, determines the computing node corresponding to each task, and encapsulates the execution relationship between tasks and computing nodes into a data packet, which is broadcast within the computing power network. Upon receiving the broadcast data packet, each computing node parses it, identifies the task it needs to execute, and begins executing the corresponding task as an execution node. After completing the task, it feeds back the task completion status to the task's source node. The task's source node, upon receiving the feedback from the task execution node, then feeds back the task completion status to the user.

[0044] The following describes a task processing method for a computing network provided in this application through specific embodiments.

[0045] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating a task processing method for a computing power network provided in an embodiment of this application. This method can be applied to management nodes in a computing power network, and its specific implementation includes:

[0046] S001. Obtain the computing power information of all computing power nodes in the computing power network.

[0047] In a computing power network, computing power nodes can obtain their own computing power information and, after obtaining their own computing power information, broadcast it within the computing power network (or propagate it separately to the computing power network management node), thereby enabling the computing power network management node to obtain the computing power information of all computing power nodes in the computing power network.

[0048] Among them, computing power information refers to information related to the computing power of computing nodes, such as the computing time rate and the number of service replicas for all applications deployed locally.

[0049] Specifically, in one embodiment of this application, obtaining the computing power information of all computing power nodes in the computing power network includes:

[0050] Step 1: Each computing node in the computing power network obtains its local computing power information and broadcasts the computing power information throughout the computing power network;

[0051] Step 2: The control node in the computing power network receives the computing power information broadcast by other computing power nodes.

[0052] That is, each computing node first obtains its own computing power information, and then broadcasts the computing power information in the computing power network, so that the control node can know the computing power information of the computing power node.

[0053] The computing power information includes computing power usage, network transmission status, and the resource requirements of received tasks.

[0054] Specifically, network transmission status is represented by a matrix. Indicates, that is It is a Q×Q matrix representing the data transmission loss between any two computing nodes; w ij Represents computing node e i With e j Loss in data transmission between them.

[0055] The computing power usage includes: Q is the number of computing power nodes; K represents the set of applications deployed in the computing power network, where K is the total number of applications.

[0056] The resource requirements of the received task: It is a Q×K matrix, p qk Let be a natural number, representing the computing node e. q Applications deployed on k The number of copies, when p qk =0 indicates that the computing power node e q Applications not deployed on k . It is a Q×K matrix, v qk A positive real number, representing the computing node e. q Applications deployed on k The smaller the value of the calculation rate, the more it indicates e. q Execute s k The higher the efficiency of the relevant task, the better if p qk =0 then v qk =+∞.

[0057] S002. Based on computing power information, linear programming algorithm and multiple constraints, make task distribution decisions for the received tasks to obtain the computing power nodes that actually execute each task.

[0058] The tasks are sent by each computing node.

[0059] The variables in the linear programming algorithm include the computation time rate and the number of service replicas for each computing node. Specifically, in the computing network, computing node e... q For each local application k Calculation rate v qk Service copy number p qk v qk This refers to computing node e q Processing applications k The time taken to process one unit of data (this unit can be byte, megabyte, gigabyte, depending on the application characteristics). qk The smaller the value, the more it indicates that e q Processing s k The higher the efficiency of the relevant tasks. qk e q Supported applications k The number of copies.

[0060] The goal of the linear programming algorithm is to complete all tasks in the task set in the shortest possible time.

[0061] Specifically, based on computing power information, linear programming algorithms, and multiple constraints, task distribution decisions are made for the received tasks to obtain the computing power nodes that actually execute each task, including:

[0062] Step 1: Obtain the computing time rate and service replica count for each computing node;

[0063] Step 2: Based on computing power information, linear programming algorithms, and multiple constraints, the problem of collaborative scheduling of computing power to process tasks is modeled as an integer linear programming problem;

[0064] Step 3: Solve the integer linear programming problem to obtain the computing power nodes that actually execute each task.

[0065] For ease of description, the three steps mentioned above will be explained in combination below.

[0066] Among them, e q v qk and p qk Quantification can be achieved through statistical processing or by directly obtaining data from the system.

[0067] Then, the problem of collaborative scheduling of computing power to process tasks is modeled as an integer linear programming problem.

[0068] After converting it into an integer linear programming problem, the computational nodes that actually execute each task can be obtained by solving the problem or by using a solution algorithm.

[0069] S003. Send each task to the corresponding computing power node so that each computing power node can process the received task.

[0070] Specifically, after obtaining the computing power nodes that actually execute each task, each task is sent to the corresponding computing power node, including:

[0071] Step 1: Encapsulate the corresponding information of each task and the computing power nodes that actually execute each task into a decision result data packet;

[0072] Step 2: Broadcast the decision result data packet in the computing power network.

[0073] That is, the corresponding information of each task and the computing power node that actually executes each task is encapsulated into a decision result data packet, and then the decision result data packet is broadcast in the computing power network.

[0074] This allows other computing nodes to parse the decision result data packet after receiving it, process the corresponding task, and obtain the processing result. In other words, other computing nodes determine the corresponding task based on the acquired decision result data packet.

[0075] Once the processing result is obtained, it can be broadcast across the computing power network. In this way, the computing power that initiated the task broadcast will receive the processing result.

[0076] Correspondingly, this computing node can also receive task processing broadcast from other computing nodes. The specific implementation process is as follows:

[0077] Step 1: Receive tasks sent by other computing power nodes and add the received tasks to the task queue;

[0078] Step 2: Execute the corresponding tasks sequentially based on the task queue to obtain the processing results.

[0079] This application provides a task processing method for a computing power network, comprising: acquiring computing power information of all computing power nodes in the computing power network; making task distribution decisions on received tasks based on computing power information, a linear programming algorithm, and multiple constraints to obtain the computing power nodes that actually execute each task; wherein, the variables of the linear programming algorithm include the computing time rate and the number of service replicas of each computing power node; and sending each task to the corresponding computing power node so that each computing power node can process the received task.

[0080] By acquiring computing power information from multiple computing nodes, and then making task distribution decisions based on this information, a linear programming algorithm, and multiple constraints, the corresponding computing node for each task is determined. Since the parameters of this linear programming algorithm include the computation time rate and the number of service replicas, which can evaluate the performance of computing nodes in the computing network for a specific application, when designing the optimal solution for collaborative task processing, only the runtime performance of the computing nodes needs to be considered, without taking into account the heterogeneous hardware configuration and software environment of the computing nodes in the computing network. This improves the efficiency of the computing network in processing tasks and increases the utilization rate of computing resources.

[0081] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating a task processing method for a computing network provided in an embodiment of this application.

[0082] In this embodiment, the method may include:

[0083] S101, each computing node in the computing power network acts as a source node to receive tasks submitted by users, and sends its own computing power information and the received task information to the computing power network management node at fixed time intervals.

[0084] Users can access computing nodes via wired or wireless means, and then encapsulate task-related information into a data packet and transmit it to the computing node. The data packet includes the application corresponding to the task and the data related to the task.

[0085] The fixed time interval for each computing power node to send information to the computing power network management node is predetermined, such as 1 minute or 10 seconds.

[0086] Each computing node sends its own computing power information to the computing power network management node, including the computing time rate and the number of service replicas for all locally deployed applications. In other words, computing node e in the computing power network... q For each local application k Calculation rate v qk Service copy number p qk v qk This refers to computing node e q Processing applications k The time taken to process one unit of data (this unit can be byte, megabyte, gigabyte, depending on the application characteristics). qk The smaller the value, the more it indicates that e q Processing s k The higher the efficiency of the relevant tasks. qk e q Supported applicationsk The number of copies. q computing power information v qk and p qk Quantification can be achieved through statistical processing or by directly obtaining data from the system.

[0087] For example, to quantize v qk First, we can statistically analyze the relationship with s. k Related historical task set H qk Then, obtain the actual response computation time and data volume of the tasks in the set. For each task in the task set, divide the actual response computation time by the data volume of that task to obtain the computation time rate of each historical task. Finally, the average or median of the computation time rates of all historical tasks is taken as v. qk The quantized value of v. For newly developed applications, since there is no historical data, v cannot be quantified using the methods described above. qk At this point, you can run the test multiple times. k The relevant tasks are identified, and the execution status of these tasks is added to the historical task set H. qk Then, the above method is used to quantify v. qk p qk It is a predefined system-level parameter that supports application parallelism.

[0088] It can be seen that by defining the calculation rate v qk Service copy number p qk When designing optimal solutions for tasks processed collaboratively using computing power, the computing power information of computing power nodes only needs to focus on the runtime performance of the nodes themselves, without considering the heterogeneous hardware configurations and software environments of the nodes in the computing power network. Furthermore, by adopting the computing power network model, when representing computing power network tasks, it is no longer necessary to predict the computational load required for the task; only the amount of data that the task needs to process needs to be obtained. This will significantly reduce the design complexity of computing power network task processing methods while ensuring the efficiency of computing power network task processing.

[0089] The task information sent by each computing node to the computing network management node includes the source node ID of each received task, the corresponding application, and the amount of related data.

[0090] S102, after the computing power network management node obtains the computing power information of each computing power node in the computing power network and the task information collected by each computing power node in the computing power network, it inputs this computing power information and task information into the pre-designed linear programming model, then uses the existing solver to calculate the task distribution strategy, determines the computing power node corresponding to each task, and then encapsulates the corresponding execution relationship between the task and the computing power node into a data packet, which is broadcast within the computing power network.

[0091] The pre-designed linear programming model will be described in detail in the next embodiment.

[0092] The solver can be an existing commercial or open-source tool, or a self-developed tool. For example, Gurobi or Cplex.

[0093] The result obtained by the solver is the task distribution strategy, that is, each task is matched with a corresponding execution computing node.

[0094] S103: After receiving the broadcast data packet, each computing node parses the data packet, determines the task it needs to perform, and starts to execute the corresponding task as an execution node. After the task is completed, it feeds back the task completion status to the source node of the task.

[0095] The process of an execution node executing a corresponding task includes first obtaining the task-related data packet from the task source node, and then starting the local application to process the task data.

[0096] S104 After receiving the feedback from the task execution node, the task source node then sends the task completion status back to the user.

[0097] The pre-designed linear programming model provided in this application will be described below through another specific embodiment.

[0098] Before explaining the model, we first formally represent the computing network and computing network tasks used in this embodiment, and at the same time, we define the symbols used in the modeling process.

[0099] The computing power network can be formally represented as: in, Let Q represent the set of computing nodes in the computing power network, where Q is the number of computing nodes. K represents the set of applications deployed in the computing power network, where K is the total number of applications. It is a Q×Q matrix, w ij Represents computing node e i With e j Loss in data transmission between them; It is a Q×K matrix, p qk Let be a natural number, representing the computing node e. q Applications deployed on k The number of copies, when p qk =0 indicates that the computing power node e q The application sk is not deployed on it. It is a Q×K matrix, v qk A positive real number, representing the computing node e. q Applications deployed on k The smaller the value of the calculation rate, the more it indicates e.q Execute s k The higher the efficiency of the relevant task, the better if p qk =0 then v qk =+∞.

[0100] In computing power networks and applications k The relevant task is formally represented as CoR. k ={R k L k F k To simplify the description of the problem, the subscript k is ignored in the following formal representation. L represents the set of computing power network tasks, where Z is the total number of tasks; L = [l zq ] Z×Q It is a Z×Q matrix, only when r z The source node is e q time l zq =1, otherwise l zq =0; F=[f z ] z It is a Z-dimensional vector, f z Let r be a real number, representing the relationship between r and r. z The size of the relevant data.

[0101] Based on the above definitions, this embodiment provides a linear programming model.

[0102] Define matrix variable X∈[x zq ] Z×Q And agreed x zq ∈{0, 1}, only when r z The execution node is e q time x zq =1, otherwise x zq =0.

[0103] Therefore, based on the above definitions, linear programming models include:

[0104] Objective function (1): min X max q∈{1，2，3，...，Q} TE q ;

[0105] Among them, equation (2):

[0106] Equation (3):

[0107] Equation (4): tttr q =max z∈{1，2，3，...，Z} x zq f z (∑ n∈[Q] l znw nq ), q∈{1, 2, 3,..., Q};

[0108] Equation (5): TE q =max(tlc q ,tttr q )+ttc q , q∈{1, 2, 3,..., Q};

[0109] Constraint (6): ∑ q∈{1，2，3，...，Q} x zq =1 z∈{1, 2, 3,..., Z};

[0110] Constraint (7): ∑ q∈{1，2，3，...，Q} x zq p qk =1 z∈{1, 2, 3,..., Z};

[0111] Constraint (8): ∑ q∈{1，2，3，...，Q} p qk >0;

[0112] Objective function (1) Minimizes all applications s within the computing power network. k The response time of the relevant tasks. Equation (2) calculates the response time of both the source node and the execution node, where e is the response time. q The execution time (tlc) required for all tasks q Equation (3) calculates that the source node is not e. q But migrating to e q The total execution time (ttc) of all tasks performed q Equation (4) calculates the data transferred from the source node to the execution node e for each task. q Required transmission time tttr q Equation (5) calculates e q The time required to complete all tasks assigned to itself (TE) q Constraint (6) ensures that all tasks will be executed, and only by one computing node. Constraint (7) ensures that the application s k The relevant tasks will only occur after the application is deployed. k The application is executed on the computing power node. Constraint (8) ensures that at least one computing power node in the computing power network has deployed the application. k .

[0113] The linear programming model established above can be solved using existing solvers or by a third-party designed algorithm, such as Gurobi or Cplex. The solver ultimately obtains the precise value of variable X, when x... zq When = 1, it is represented as r zAssign execution node e q In a computing power network, any task r z The execution node is calculated as follows:

[0114]

[0115] work out After obtaining the corresponding specific values, then, the computing power nodes The computing power network is based on a linear programming model for task r. z The matched execution node. For example, if Then e1 represents the computing power network based on the linear programming model, and r represents the task. z Matching execution nodes.

[0116] The task processing apparatus for the computing power network provided in the embodiments of this application is described below. The task processing apparatus for the computing power network described below and the task processing method for the computing power network described above can be referred to in correspondence.

[0117] Please refer to Figure 3 , Figure 3 This is a schematic diagram of the structure of a task processing device for a computing network provided in an embodiment of this application.

[0118] In this embodiment, the device may include:

[0119] The computing power information acquisition module 001 is used to acquire computing power information of all computing power nodes in the computing power network;

[0120] The task planning module 002 is used to make task distribution decisions based on computing power information, linear programming algorithm and multiple constraints to obtain the computing power nodes that actually execute each task; among them, the variables of the linear programming algorithm include the computing time rate and the number of service replicas of each computing power node.

[0121] The task forwarding module 003 is used to send each task to the corresponding computing power node so that each computing power node can process the received task.

[0122] In one specific embodiment of this application, the computing power information acquisition module 001 is specifically used to acquire local computing power information and broadcast the computing power information in the computing power network;

[0123] It receives computing power information broadcast by other computing power nodes.

[0124] In one specific embodiment of this application, the computing power information includes computing power usage, network transmission status, and the resource requirements of received tasks.

[0125] In one specific embodiment of this application, the planning objective of the linear programming algorithm is to complete all tasks within the task set in the shortest possible time.

[0126] In one specific embodiment of this application, the task forwarding module 003 is specifically used to encapsulate the corresponding information of each task and the computing power node that actually executes each task into a decision result data packet after obtaining the computing power node that actually executes each task.

[0127] The decision result data packets are broadcast in the computing power network.

[0128] In one specific embodiment of this application, it further includes:

[0129] The task processing module is used by other computing nodes to determine the corresponding tasks based on the acquired decision result data packets.

[0130] In one specific embodiment of this application, it further includes:

[0131] The task processing module is used to receive tasks sent by other computing power nodes and add the received tasks to the task queue;

[0132] The corresponding tasks are executed sequentially based on the task queue to obtain the processing results.

[0133] In one specific embodiment of this application, the task planning module 002 is specifically used to obtain the computing time rate and service replica number of each computing power node; based on computing power information, linear programming algorithm and multiple constraints, the problem of computing power collaborative scheduling and processing tasks is modeled as an integer linear programming problem; the integer linear programming problem is solved to obtain the computing power nodes that actually execute each task.

[0134] This application also provides a computing device, please refer to... Figure 4 , Figure 4 This application provides a schematic diagram of the structure of a computing device, which may include:

[0135] Memory, used to store computer programs;

[0136] A processor is used to execute computer programs and implement the steps of any of the task processing methods of a computing network as described above.

[0137] like Figure 4 The diagram shows the structural composition of a computing device, which may include a processor 10, a memory 11, a communication interface 12, and a communication bus 13. The processor 10, memory 11, and communication interface 12 all communicate with each other through the communication bus 13.

[0138] In this embodiment, the processor 10 may be a central processing unit (CPU), an application-specific integrated circuit, a digital signal processor, a field-programmable gate array, or other programmable logic devices.

[0139] The processor 10 can call the program stored in the memory 11. Specifically, the processor 10 can execute the operations in the embodiment of the abnormal IP identification method.

[0140] The memory 11 is used to store one or more programs. The programs may include program code, which includes computer operation instructions. In this embodiment, the memory 11 stores at least a program for implementing the following functions:

[0141] The computing power node obtains computing power information from multiple corresponding nodes in the computing power network;

[0142] Based on the computing power information of multiple nodes, a linear programming algorithm, and multiple constraints, a task distribution decision is made for the received tasks to obtain the computing power nodes that actually execute each task; among them, the variables of the linear programming algorithm include the computing time rate and the number of service replicas of each computing power node.

[0143] Each task is sent to the corresponding computing node so that the computing node can process the received task.

[0144] In one possible implementation, the memory 11 may include a program storage area and a data storage area, wherein the program storage area may store the operating system and applications required for at least one function; and the data storage area may store data created during use.

[0145] In addition, memory 11 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device or other volatile solid-state storage device.

[0146] Communication interface 12 can be an interface for the communication module, used to connect with other devices or systems.

[0147] Of course, it should be noted that, Figure 4 The structure shown does not constitute a limitation on the computing device in the embodiments of this application. In practical applications, the computing device may include more than Figure 4 More or fewer components as shown, or combinations of certain components.

[0148] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement the steps of any of the above-described task processing methods for computing power networks.

[0149] The computer-readable storage medium may include 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.

[0150] For a description of the computer-readable storage medium provided in this application, please refer to the above method embodiments; further details will not be repeated here.

[0151] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0152] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0153] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0154] The foregoing has provided a detailed description of the task processing method, task processing apparatus, computing device, and computer-readable storage medium for a computing power network provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and core ideas of this application. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of the claims of this application.

Claims

1. A task processing method for a computing power network, characterized in that, include: Obtain computing power information for all computing power nodes in the computing power network; Based on the computing power information, linear programming algorithm, and multiple constraints, a task distribution decision is made for the received tasks to obtain the computing power nodes that actually execute each task; wherein, the variables of the linear programming algorithm include the computing time rate and the number of service replicas of each computing power node; the computing time rate is the time spent by the computing power node to process one unit of data of the application; Each of the tasks is sent to the corresponding computing power node so that each computing power node can process the received tasks; Specifically, based on the computing power information, linear programming algorithm, and multiple constraints, a task distribution decision is made for the received tasks to obtain the computing power nodes that actually execute each task, including: Obtain the computing time rate and service replica count for each computing node; Based on the computing power information, the linear programming algorithm, and the multiple constraints, the problem of computing power collaborative scheduling and processing tasks is modeled as an integer linear programming problem; Solve the integer linear programming problem to obtain the computing nodes that actually execute each task.

2. The task processing method according to claim 1, characterized in that, The computing power information includes computing power usage, network transmission status, and the resource requirements of received tasks.

3. The task processing method according to claim 1, characterized by, The goal of the linear programming algorithm is to complete all tasks in the task set in the shortest possible time.

4. The task processing method of claim 1, wherein, Sending each of the tasks to the corresponding computing power nodes includes: The corresponding information of each task and the computing power nodes that actually execute each task is encapsulated into a decision result data packet; The decision result data packet is broadcast in the computing power network.

5. The task processing method according to claim 4, characterized by, Also includes: Other computing nodes determine the corresponding tasks based on the obtained decision result data packets.

6. The task processing method of claim 1, wherein, Also includes: Receive tasks sent by other computing power nodes and add the received tasks to the task queue; The corresponding tasks are executed sequentially based on the task queue to obtain the processing results.

7. A task processing device for a computing network, characterized in that, include: The computing power information acquisition module is used to acquire computing power information of all computing power nodes in the computing power network; The task planning module is used to make task distribution decisions based on the computing power information, linear programming algorithm, and multiple constraints to obtain the computing power nodes that actually execute each task; wherein, the variables of the linear programming algorithm include the computing time rate and the number of service replicas of each computing power node; the computing time rate is the time spent by the computing power node to process one unit of data of the application; The task forwarding module is used to send each of the tasks to the corresponding computing power nodes so that each computing power node can process the received tasks. Specifically, the task planning module is used to obtain the computing time rate and service replica count of each computing power node; based on the computing power information, the linear programming algorithm, and multiple constraints, to model the problem of computing power collaborative scheduling and processing tasks as an integer linear programming problem; and to solve the integer linear programming problem to obtain the computing power nodes that actually execute each task.

8. A computing device, comprising: include: Memory, used to store computer programs; A processor, configured to implement the steps of the task processing method as described in any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the steps of the task processing method according to any one of claims 1 to 6.

Images