Wide-area scheduling method, device and system for task requests, medium and program product

Abstract

The embodiment of the invention provides a wide-area scheduling method, device and system for task requests, a medium and a program product, and relates to the technical field of task scheduling. The method comprises the following steps: receiving a data processing task request; analyzing the task request to obtain an inter-task dependency relationship of at least one task corresponding to the task request and a demand of each task; obtaining resource state data, wherein the resource state data comprises a global data resource view and a global physical resource view; formulating a calculation task scheduling strategy considering data locality and a single collaborative flow transmission scheduling strategy according to the dependency relationship between the tasks and the demand and resource state data of each task; and according to the initial bandwidth and the minimum communication time of all the collaborative flows, transmission rescheduling optimization of the multiple collaborative flows is executed, and all the collaborative flows comprise the collaborative flows which are currently executed and the collaborative flows corresponding to the task request. According to the embodiment of the invention, the data task processing performance and the multi-dimensional resource utilization rate are improved.

Description

Technical Field

[0001] The present application relates to the technical field of task scheduling, and in particular to a method, device, system, medium, and program product for wide-area scheduling of task requests. Background Art

[0002] As scientific and engineering computational problems become increasingly complex, big data applications urgently require the coordinated utilization of distributed computing resources in wide-area environments to support efficient computing. As a critical infrastructure supporting computing interconnection and data flow, wide-area distributed computing environments have attracted widespread attention due to their ability to provide large-scale computing and storage resources and leverage the aggregation effect of widely distributed resources. In wide-area distributed computing environments, task scheduling, through task characteristic analysis and global resource management, selects target nodes for each task, which is crucial for efficient system operation. However, the diversity of tasks and resources, the geographical dispersion of resources, and the significant communication costs in wide-area networks increase the complexity of wide-area scheduling. Furthermore, existing wide-area scheduling solutions focus on orchestrating computing resources while lacking consideration for the linkage between data and communication resources, making it difficult to achieve multi-dimensional resource collaborative optimization. The lack of joint consideration of data transmission and network scheduling can lead to underutilized resources and poor workload performance. Therefore, designing a wide-area collaborative scheduling method that considers the linkage between computing tasks, data storage, and network communication to achieve multi-dimensional resource collaborative optimization in wide-area distributed computing environments has become a key technical challenge that needs to be addressed in the current wide-area scheduling field.

[0003] Currently, mainstream methods for wide-area scheduling include computing task scheduling and communication scheduling, which optimize task computation and data transmission, respectively. Computing task scheduling technology schedules tasks to heterogeneous computing nodes by determining when, where, and how to offload them. Traditional computing task scheduling methods often exhibit problems such as inefficiency and poor adaptability when faced with dynamic task demands and multi-dimensional heterogeneous resource management. In wide-area distributed environments, the high data communication costs, transmission delays, and complex dependencies between tasks of wide-area networks are particularly prominent. Communication scheduling technology optimizes network-layer data flow routing and bandwidth allocation to improve bandwidth utilization and reduce data transmission time.

[0004] In a wide-area distributed environment, since the distribution of data and tasks is more extensive and complex, the concept and technology based on coflow scheduling can be used to better manage and coordinate large amounts of data transmission. Coflow scheduling aims to simultaneously consider the data transmission requirements between multiple related tasks to optimize the overall data transmission efficiency and task completion time. Traditional Coflow scheduling technology is mainly based on bandwidth allocation to improve network bandwidth utilization and reduce coflow completion time (CCT). In addition, it also considers the network topology to better plan the transmission path of data flow and reduce network congestion. However, traditional Coflow scheduling methods face the problems of low scheduling efficiency and poor adaptability in dynamic and changing distributed environments. Moreover, in practical applications, it is difficult for existing Coflow scheduling methods to accurately obtain prior information about task requirements, data flow, network topology and resource availability, such as real-time occupancy of network bandwidth, network congestion, server load, data flow size and importance, resulting in poor Coflow scheduling effect. The limitations of the above methods provide a prerequisite for the design of this patent.

[0005] In the field of wide-area scheduling, the related methods of existing technologies mainly have the following problems:

[0006] Computational task scheduling and data flow communication scheduling are independent of each other. Existing wide-area scheduling technologies focus on orchestrating computing resources without considering the integration with data transmission and network scheduling. This makes it difficult to achieve multi-dimensional resource coordination and optimization, which can lead to underutilized resources and poor workload performance. While data locality-based computing task scheduling methods can assign tasks to appropriate computing nodes while reducing the amount of data transmitted, data locality scheduling does not necessarily minimize data transmission time. This is because when computing nodes use the network without any coordination, network bandwidth utilization can be very poor. On the other hand, while Coflow co-flow scheduling methods improve network bandwidth utilization to accelerate data transmission, existing Coflow co-flow scheduling methods focus on optimizing communication time given the source and destination nodes of each data flow. This decouples coflow scheduling from data locality scheduling. In this case, data transmission time achieved with Coflow scheduling is suboptimal, not optimal. This is because the associated network-layer optimizations fail to recognize that Coflow scheduling may affect the final communication time. Summary of the Invention

[0007] At least one embodiment of the present application provides a method, apparatus, system, medium, and program product for wide-area scheduling of task requests, for improving data task processing performance and multi-dimensional resource utilization.

[0008] In order to solve the above technical problems, this application is implemented as follows:

[0009] In a first aspect, an embodiment of the present application provides a wide-area scheduling method for task requests, including:

[0010] Receive task requests for data processing;

[0011] Parsing the task request to obtain inter-task dependencies and requirements of each task of at least one task corresponding to the data request;

[0012] Acquiring resource status data, the resource status data including a global data resource view and a global physical resource view;

[0013] Formulate a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that consider data locality based on the dependencies between the tasks and the requirements and resource status data of each task; wherein the computing task scheduling strategy includes the execution nodes of each task, and the single collaborative flow transmission scheduling strategy includes the initial bandwidth of the data flow between the nodes and the minimum communication time of the collaborative flow;

[0014] Transmission rescheduling optimization of multiple collaborative flows is performed according to initial bandwidths and minimum communication times of all collaborative flows, wherein all collaborative flows include the currently executing collaborative flow and the collaborative flow corresponding to the task request.

[0015] Optionally, based on the dependencies between the tasks and the requirements and resource status data of each task, a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that consider data locality are formulated, including:

[0016] Determining a collaborative flow according to the dependency relationship between the tasks, where the collaborative flow includes at least one data flow;

[0017] According to the requirements of each task and the resource status data, a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy considering data locality are determined using a preset algorithm.

[0018] Optionally, based on the initial bandwidth and minimum communication time of all co-flows, transmission rescheduling optimization of multiple co-flows is performed, including:

[0019] Determining a weight factor of each collaborative flow according to the minimum communication time of all collaborative flows;

[0020] Using the weight factor to weight the initial bandwidth of the data flow of each collaborative flow to obtain the updated bandwidth of each data flow; wherein the weight factor is positively correlated with the minimum communication time;

[0021] According to the updated bandwidth of each data flow, transmission scheduling of multiple coordinated flows is performed.

[0022] Optionally, the preset algorithm is a meta-heuristic algorithm.

[0023] Optionally, a preset algorithm is used to determine a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that considers data locality, including:

[0024] Establishing a model for an optimization problem, the model including an objective function and constraints, wherein the objective function is the communication time of a data flow from node i to node j in the collaborative flow, and the communication time is the ratio of the data volume of the data flow to the bandwidth of the data flow; the constraints include at least one of a first constraint, a second constraint, and a third constraint, wherein the first constraint is a constraint on the data volume of the data flow transmitted from node i to node j, the second constraint is a constraint that the bandwidth allocated to the data flow from node i to node j is less than the available bandwidth from node i to node j, and the third constraint is a constraint that the same data partition can only be allocated to one node;

[0025] The model of the optimization problem is solved by using a meta-heuristic algorithm to obtain the computing task scheduling strategy and the single collaborative flow transmission scheduling strategy.

[0026] Optionally, also include:

[0027] After the task request is processed, the access popularity of the data used in this task request is collected;

[0028] The layout of the data used in the task request on each node is adjusted according to the popularity of the data, the computing power and / or the load of each node.

[0029] In a second aspect, an embodiment of the present application provides a wide-area scheduling device for task requests, including:

[0030] A task parsing module is configured to receive a data processing task request; parse the task request to obtain at least one task corresponding to the data request, attribute data of each task, dependencies between tasks, and requirements of each task; the attribute data may include at least one of the following: task type, task scale, task priority, timeliness requirement, and data source;

[0031] Data management module, used to obtain a global view of data resources;

[0032] Resource management module, used to obtain a global view of physical resources;

[0033] A general computing integrated scheduler is used to formulate a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that considers data locality based on the dependencies between the tasks and the requirements and resource status data of each task; wherein the resource status data includes a global data resource view and a global physical resource view; the computing task scheduling strategy includes the execution nodes of each task, and the single collaborative flow transmission scheduling strategy includes the initial bandwidth of the data flow between nodes and the minimum communication time of the collaborative flow; based on the initial bandwidth and minimum communication time of all collaborative flows, transmission rescheduling optimization of multiple collaborative flows is performed, wherein all collaborative flows include the currently executing collaborative flow and the collaborative flow corresponding to the task request.

[0034] Optionally, the integrated computing scheduler is further configured to:

[0035] Determining a collaborative flow according to the dependency relationship between the tasks, where the collaborative flow includes at least one data flow;

[0036] According to the requirements of each task and the resource status data, a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy considering data locality are determined using a preset algorithm.

[0037] Optionally, the integrated computing scheduler is further configured to:

[0038] Determining a weight factor of each collaborative flow according to the minimum communication time of all collaborative flows;

[0039] Using the weight factor to weight the initial bandwidth of the data flow of each collaborative flow to obtain the updated bandwidth of each data flow; wherein the weight factor is positively correlated with the minimum communication time;

[0040] According to the updated bandwidth of each data flow, transmission scheduling of multiple coordinated flows is performed.

[0041] Optionally, the preset algorithm is a meta-heuristic algorithm.

[0042] Optionally, a preset algorithm is used to determine a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that considers data locality, including:

[0043] Establishing a model for an optimization problem, the model including an objective function and constraints, wherein the objective function is the communication time of a data flow from node i to node j in the collaborative flow, and the communication time is the ratio of the data volume of the data flow to the bandwidth of the data flow; the constraints include at least one of a first constraint, a second constraint, and a third constraint, wherein the first constraint is a constraint on the data volume of the data flow transmitted from node i to node j, the second constraint is a constraint that the bandwidth allocated to the data flow from node i to node j is less than the available bandwidth from node i to node j, and the third constraint is a constraint that the same data partition can only be allocated to one node;

[0044] The model of the optimization problem is solved by using a meta-heuristic algorithm to obtain the computing task scheduling strategy and the single collaborative flow transmission scheduling strategy.

[0045] Optionally, the integrated computing scheduler is further configured to:

[0046] After the task request is processed, the access popularity of the data used in this task request is collected;

[0047] The layout of the data used in the task request on each node is adjusted according to the popularity of the data, the computing power and / or the load of each node.

[0048] In a third aspect, an embodiment of the present application provides a wide-area scheduling system for task requests, comprising: a processor, a memory, and a program stored on the memory and executable on the processor, wherein the program, when executed by the processor, implements the steps of the method described in the first aspect.

[0049] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a program is stored. When the program is executed by a processor, the steps of the method described in any one of the first aspects are implemented.

[0050] In a fifth aspect, an embodiment of the present application provides a computer program product, comprising computer instructions, which, when executed by a processor, implement the steps of the method described in any one of the first aspects.

[0051] Compared with the prior art, the wide-area scheduling method, device, system, medium, and program product for task requests provided by the embodiments of the present application formulate a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that considers data locality based on the dependencies between tasks and the requirements and resource status data of each task, and perform transmission rescheduling optimization of multiple collaborative flows based on the initial bandwidth and minimum communication time of all collaborative flows, thereby improving the processing performance of data tasks and multi-dimensional resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present application. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0053] Figure 1 A schematic diagram of a wide-area scheduling device for task requests according to an embodiment of the present application;

[0054] Figure 2 A flow chart of a wide-area scheduling method for task requests according to an embodiment of the present application;

[0055] Figure 3 Another flow chart of the wide-area scheduling method for task requests according to an embodiment of the present application;

[0056] Figure 4 This is one of the scheduling example diagrams of the related technology;

[0057] Figure 5 This is the second scheduling example diagram of the related technology;

[0058] Figure 6 This is a schematic diagram of the scheduling process of multiple collaborative flows according to an embodiment of the present application;

[0059] Figure 7 This is an example diagram of scheduling multiple collaborative flows according to an embodiment of the present application;

[0060] Figure 8 A schematic diagram of a wide-area scheduling system for task requests according to an embodiment of the present application. DETAILED DESCRIPTION

[0061] The terms "first", "second", etc. in this application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way are interchangeable where appropriate, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same type, and do not limit the number of objects, for example, the first object can be one or more. In addition, the term "and / or" in this application describes the association relationship of associated objects, indicating that three relationships can exist, for example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects associated before and after are in an "or" relationship.

[0062] The term "indication" in this application can be either a direct indication (or explicit indication) or an indirect indication (or implicit indication). A direct indication can be understood as the sender explicitly informing the receiver of specific information, the operation to be performed, or the requested result, etc. in the instruction sent; an indirect indication can be understood as the receiver determining the corresponding information based on the instruction sent by the sender, or making a judgment and determining the operation to be performed or the requested result, etc. based on the judgment result.

[0063] Please refer to Figure 1 An embodiment of the present application provides a wide-area scheduling device for task requests, including a cloud computing cluster, an edge computing cluster, a programmable switch cluster, a task parsing module, a data management module, a resource management module, and a general computing (communication computing) integrated scheduling module.

[0064] The cloud computing cluster comprises multiple high-performance cloud computing nodes, each interconnected via a high-speed network, enabling data and task sharing and collaboration. As a powerful back-end computing resource pool, the cloud computing cluster boasts exceptional computing power and vast storage capacity, handling large-scale, complex computing tasks and storing massive amounts of data.

[0065] The edge computing cluster consists of multiple geographically distributed edge computing nodes. These nodes have strong local computing and storage capabilities, enabling rapid response to local data processing needs. The edge computing cluster performs computations at or near the data source, reducing data transmission latency and bandwidth usage.

[0066] The programmable switch cluster comprises multiple high-performance, programmable switches, leveraging software-defined networking (SDN) technology to achieve refined control of network traffic and resources. By flexibly configuring network traffic, the programmable switch cluster optimizes data transmission paths and network resource utilization. This ensures efficient and reliable data transmission between cloud computing and edge computing clusters, and supports dynamic adjustment of network policies to meet the needs of different tasks.

[0067] The task parsing module is responsible for performing in-depth semantic and syntactic analysis of received tasks, accurately extracting task attribute data. This attribute data includes task type (such as data mining, model training, or real-time data analysis), task scale (such as the amount of data involved and computational complexity), task priority, task timeliness requirements, and data source. Accurately acquiring this attribute data provides a solid foundation for subsequent resource assessment and scheduling strategy development.

[0068] The data management module is responsible for unified management of data resources across the entire big data platform (i.e., the wide-area scheduling device). By aggregating and sorting data resources distributed across different nodes and storage devices at the local, regional, and wide-area levels, it establishes a unified data directory and index, providing a global view of data resources to support aggregated management and unified access. This enables a clear understanding of data distribution and availability during the scheduling process, helping to more accurately formulate scheduling strategies, reduce data search and acquisition time, and improve overall data processing efficiency.

[0069] The resource management module is responsible for comprehensive monitoring and meticulous management of the computing, storage, and network resources of the entire big data platform (i.e., the wide-area scheduling device). It provides real-time visibility into resource usage and availability, ensuring efficient resource utilization and dynamic resource adjustments. It allocates and schedules resources based on task requirements and current resource status.

[0070] The integrated computing scheduling module organically integrates data locality-sensitive computing task scheduling with collaborative data flow scheduling. It is responsible for comprehensively scheduling computing tasks for cloud computing clusters and edge computing clusters, as well as communication tasks for programmable switch clusters, using compute controllers and network controllers. It coordinates computing task execution, data routing, and data flow bandwidth allocation. The integrated computing scheduling module includes a scheduling engine and a scheduling policy library. The scheduling engine, composed of a compute controller and a network controller, is responsible for executing scheduling decisions in real time. The policy library stores various strategies and algorithms for optimizing scheduling.

[0071] Please refer to Figure 2 , an embodiment of the present application provides a wide-area scheduling method for task requests, comprising the following steps:

[0072] Step 21: Receive a data processing task request.

[0073] Step 22: parse the task request to obtain inter-task dependencies and requirements of each task of at least one task corresponding to the task request.

[0074] Here, the task parsing module can be used to perform semantic and grammatical analysis on the task request to determine at least one task corresponding to the task request and the attribute data of each task. These attribute data include but are not limited to the type of task (such as data mining tasks, model training tasks, real-time data analysis tasks, etc.), the scale of the task (such as the amount of data involved in the task, the computational complexity, etc.), the task priority, the timeliness requirements of the task, the source of the data, etc. Then, based on the attribute data of each task, the dependencies between tasks and the requirements of each task are determined. The dependencies between tasks usually represent the execution order between tasks. For example, when the execution of task B requires the use of the output results of task A, task B depends on task A. The requirements of the task can be task priority, timeliness requirements for task execution (for example, how long it takes to complete the execution), etc.

[0075] Step 23: Acquire resource status data, where the resource status data includes a global data resource view and a global physical resource view.

[0076] Here, the data management module can obtain a global data resource view, which includes the data resources of each node (cloud computing nodes, edge computing nodes, etc.). The resource management module can obtain a global physical resource view, which includes the usage and availability of physical resources such as computing resources, storage resources, and network resources of each node (cloud computing nodes, edge computing nodes, etc.).

[0077] Step 24: Based on the dependencies between the tasks and the requirements and resource status data of each task, formulate a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that considers data locality; wherein the computing task scheduling strategy includes the execution nodes of each task, and the single collaborative flow transmission scheduling strategy includes the initial bandwidth of the data flow between the nodes and the minimum communication time of the collaborative flow.

[0078] Here, a collaborative flow is determined based on the dependencies between the tasks, and the collaborative flow includes at least one data flow. Then, based on the requirements of each task and the resource status data, a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that considers data locality are determined using a preset algorithm.

[0079] Step 25 : performing transmission rescheduling optimization of multiple collaborative flows according to the initial bandwidth and minimum communication time of all collaborative flows, wherein all collaborative flows include the currently executing collaborative flow and the collaborative flow corresponding to the task request.

[0080] Since there may be other collaborative flows being executed at this time, after the embodiment of the present application calculates the computing task scheduling strategy and collaborative flow transmission scheduling strategy of the single collaborative flow in step 24, it further obtains the initial bandwidth and minimum communication time of the collaborative flow being executed. Then, based on the collaborative flow being executed and the minimum communication time of the collaborative flow calculated in step 24, the weight factor of each collaborative flow is determined, and the initial bandwidth of the data flow of each collaborative flow is weighted using the weight factor to obtain the updated bandwidth of each data flow; wherein the weight factor is positively correlated with the minimum communication time. Then, based on the computing task scheduling strategy of each collaborative flow and the updated bandwidth of each data flow, the transmission scheduling of multiple collaborative flows is performed.

[0081] Through the above steps, the embodiment of the present application constructs a joint optimization problem of computing task scheduling and collaborative job data flow scheduling considering data locality, and optimizes the execution of distributed data operators by combining the two perspectives of reducing data transmission volume and optimizing data flow network transmission, thereby reducing the processing time of data tasks and improving the utilization of various resources.

[0082] Here, the preset algorithm may specifically be a meta-heuristic algorithm.

[0083] In the embodiment of the present application, a preset algorithm is used to determine the implementation of a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that considers data locality, specifically including:

[0084] (1) Establishing a model for the optimization problem, the model including an objective function and constraints, wherein the objective function is the communication time of the data flow from node i to node j in the collaborative flow, and the communication time is the ratio of the data volume of the data flow to the bandwidth of the data flow; the constraints include at least one of a first constraint, a second constraint, and a third constraint, wherein the first constraint is a constraint on the data volume of the data flow transmitted from node i to node j, the second constraint is a constraint that the bandwidth allocated to the data flow from node i to node j is less than the available bandwidth from node i to node j, and the third constraint is a constraint that the same data partition can only be allocated to one node;

[0085] (2) Using a meta-heuristic algorithm, the model of the optimization problem is solved to obtain the computing task scheduling strategy and the single collaborative flow transmission scheduling strategy.

[0086] Optionally, after the task request is processed, the embodiment of the present application collects the access popularity of the data used in the task request. Then, based on the popularity of the data, the computing power and / or load of each node, the layout of the data used in the task request on each node is adjusted.

[0087] Another embodiment of the present application provides a wide area scheduling method for task requests, such as Figure 3 As shown, the following steps are included:

[0088] The general computing integrated scheduler receives task requests for data processing;

[0089] Get the global data resource view and the global physical resource view;

[0090] Parse the task request to extract the attribute data of the task;

[0091] Comprehensively consider task requirements, inter-task dependencies, global data distribution, and computing and communication resource status to formulate computing task scheduling strategies and collaborative flow transmission scheduling strategies that take data locality into consideration, ensuring the coordinated optimization of computing task scheduling and data communication transmission scheduling;

[0092] The general computing integrated scheduler sends a task scheduling request to the computing controller according to the task scheduling strategy;

[0093] The computing controller issues task scheduling instructions to the cloud computing cluster and the edge computing cluster according to the task scheduling request to execute the task scheduling process;

[0094] The integrated computing scheduler sends a collaborative flow scheduling request to the network controller according to the collaborative flow transmission scheduling strategy;

[0095] The network controller sends a collaborative flow scheduling instruction (eg, flow table information) to the programmable switch cluster according to the collaborative flow scheduling request.

[0096] In this way, the cloud computing cluster and edge computing cluster perform data processing tasks, and the programmable switch cluster responds to data forwarding operations based on flow table information. After the job is completed, the access popularity of the data used in the calculation is collected. Taking into account the computing power and / or load of each data center node, the layout of the data used in the task request is adjusted on each node to reduce the amount of data transmitted during subsequent task execution.

[0097] The scheduling strategy is formulated through the following steps. The embodiment of the present application mathematically constructs a joint optimization problem of computing task scheduling and collaborative job data flow scheduling considering data locality. By combining the two perspectives of reducing data transmission volume and optimizing data flow network transmission, the distributed data operator execution is optimized, thereby reducing data task processing time. To solve this problem, the embodiment of the present application designs an online scheduling method based on a metaheuristic algorithm. By utilizing the powerful ability of the metaheuristic algorithm in searching for the global optimal solution, the scheduling strategy is dynamically adjusted according to the real-time changing resource status and task requirements to achieve efficient task execution and optimal resource allocation.

[0098] Computational task scheduling that considers data locality determines how tasks within a job, consisting of a set of tasks, are allocated near the data and how the data they depend on is retrieved. This generates a data scheduling plan, including the node destinations for data tuples to be transmitted over the network. Data locality refers to the tendency for computational tasks to access data that is physically nearby, rather than data that requires long-distance network transmission. This scheduling approach can significantly reduce data transmission overhead and latency, thereby improving computational efficiency. Figure 4 A diagram of computing task scheduling that considers data locality for related technologies shows three possible scheduling plans for aggregation on a 3-node system. Each data tuple is represented by its key, and the superscript of each key represents the frequency of the data tuple. For example, 4 2 Indicates that there are 2 tuples with keys 4, 4 2 is the data block on node 1, and 4 2 and 4 5 is the input data partition. In addition, the dotted arrows represent the scheduling results, i.e., the target node of each tuple. Figure 4 (a) shows the data locality schedule P1 based on the hash algorithm. The node to which the data tuple should be assigned is determined by the hash value. The hash value of the key is calculated as the modulus of the key value and the number of nodes plus 1. The amount of data transmitted is quantified by the number of data tuples moved to the remote node. The communication cost of this schedule is 12. Figure 4 (b) and (c) in Fig. 3 show the suboptimal data locality schedule P2 and the optimal data locality schedule P3, with communication costs of 11 and 8 respectively.

[0099] In a wide-area distributed environment, since the distribution of data and tasks is more extensive and complex, technology based on Coflow collaborative flow scheduling can be used to better manage large amounts of data transmission. Coflow scheduling aims to simultaneously consider the data transmission needs between multiple related tasks to optimize the overall data transmission efficiency and task completion time. Coflow scheduling does not schedule a single data flow, but more comprehensively considers the interrelationships and dependencies between multiple concurrent data transmission flows, focusing on minimizing the completion time of the slowest flow in the job. The Coflow abstraction is used to define a set of parallel data flows that are related to each other and also share a common performance goal. In order to optimize the data communication time for big data task processing, it is necessary to optimize the communication scheduling at the collaborative flow level rather than the individual flow level. This is because the data communication time depends on the Coflow completion time, rather than the time to complete a single data flow in the Coflow.

[0100] A single flow in a Coflow schedule can be defined by a tuple of the form [src,des,d], where src and des are the source and destination nodes, and d is the flow size. For example, Figure 4The data flow generated by the schedule in (c) can be viewed as a coflow with three independent data flows. When keys 4 and 5 move from node 1 to node 2, a single data flow will be generated in the actual execution because their source and destination are the same. In this case, Coflow scheduling can improve the communication time of the parallel data flows generated in the distributed data operator. Figure 5 A schematic diagram of Coflow scheduling, where the vertical height of the filled bar represents the bandwidth.

[0101] The integration of the two methods can reduce the data communication time of data task processing. Computational task scheduling considering data locality can reduce the amount of transmitted data, and Coflow scheduling can optimize network communication and increase the data transmission rate. However, the communication time in this scheme is not optimal. The main reasons include: (1) Although computing task scheduling considering data locality plays a certain role in reducing the amount of transmitted data, it lacks awareness of communication time and network status. (2) Coflow scheduling assumes that the detailed information of each data flow [src, des, v] is known before Coflow starts. The related network layer optimization does not realize that Coflow scheduling may affect the final communication time. For example, Figure 5 Figures (a) and (b) show that when using optimal Coflow scheduling, the suboptimal data locality schedule P2 even uses less network communication time than the optimal data locality schedule P3. This means that the scheduling engine should choose the suboptimal data locality schedule P2 for execution. This decision needs to be adjusted dynamically when the available bandwidth between nodes changes, for example, when that bandwidth is consumed by other services. Figure 5 (c) shows that when the bandwidth is halved, schedule P3 has less network communication time than P2. In this case, the scheduling engine should select the optimal data locality schedule P3 for execution. Therefore, to minimize the data communication time for big data task processing, data locality scheduling should be considered together with Coflow scheduling.

[0102] As can be seen from the above examples, by jointly considering data locality in computing task scheduling and Coflow scheduling, data processing and data transmission can be effectively combined, thereby minimizing the processing time of large data tasks. Next, the implementation method of wide-area scheduling considering data locality and Coflow transmission described in the embodiments of this application is explained in detail.

[0103] Modeling the optimization problem of jointly considering computing task scheduling and Coflow scheduling with data locality: The goal of wide-area scheduling considering data locality and Coflow transmission is to minimize the data communication time of task processing by determining the data locality scheduling for computing tasks and the communication scheduling for data transmission. According to the given data locality scheduling plan, all information of each data flow can be obtained, including the data source, destination node, data flow size and communication link resource status. In this case, the problem of minimizing the data communication time T of large data task processing can be expressed as minimizing the collaborative flow completion time CCT composed of all individual data flows. Since the communication time of each data flow is the same in the optimal Coflow scheduling, the communication time can be calculated by dividing the data flow size by the bandwidth allocated to its communication link. Therefore, the optimization problem of jointly considering computing task scheduling and Coflow scheduling with data locality can be modeled as:

[0104]

[0105] Among them, d i,j Indicates the size of the data flow generated by the movement of data from one node i to another node j. j,k ∈{0, 1} indicates whether data partition k is assigned to node j. If it is assigned to node j, the value is 1, otherwise it is 0. i,j Represents the bandwidth allocated for data streaming. N represents the number of nodes, and the tuples on each node are divided into k parts. In general, the partitioning of data is based on the hash value of the key, the hash value of the nth part is denoted as n, and the data size at node i is denoted as s i,n Constraint C1 indicates that the data flow size is calculated by adding the sizes of all data blocks transmitted from node i to node j. Constraint C2 indicates that the link bandwidth consumption should be less than the available bandwidth, where L i,j represents the path from node i to node j; l represents a link from node i to node j; the channel may include one or more links; B l represents the bandwidth of the link. Constraint C3 represents the constraint of data partitioning on the target node. The problem is a mixed integer nonlinear optimization problem, which is proved to be NP-hard. It is difficult to solve. However, for a given data partitioning method, the available bandwidth r of each node i is i For distributed data operators, it is a constant that can be collected through resource status or allocated by the network controller, and each data block s i,n The size of will also be constant. Therefore, the optimization problem can be transformed into a mixed integer linear programming problem.

[0106] To solve this problem, the present invention designs an online scheduling method based on a metaheuristic algorithm. By leveraging the powerful ability of the metaheuristic algorithm in searching for the global optimal solution, the scheduling strategy is dynamically adjusted according to the real-time changing resource status and task requirements to achieve efficient task execution and optimal resource allocation. To fit the metaheuristic, the optimization problem of jointly considering data locality in computing task scheduling and Coflow scheduling can be rewritten as:

[0107]

[0108] Among them, F is the fitness function, S i and R j Respectively represent the communication time for each node to send and receive data.

[0109] The online scheduling process based on the metaheuristic algorithm is as follows. The general integrated scheduler obtains task requirements, task dependencies, data flows, network topology, and resource status information. Then, the position of each search agent is randomly generated and each position is converted into an initial solution to the optimization problem. After that, the fitness value of each initial solution is calculated according to the equation, and the X with the minimum value is stored. * As the best search agent. Subsequently, an iterative process is started to optimize the optimal solution. In each iteration, the position of each search agent is updated according to the rules of surrounding, shrinking and spiraling. Then, the updated position is converted into a new solution to calculate the fitness value in the current iteration. The current optimal solution X * The new solution and the old solution will be selected for update according to the fitness value and used to update the position of the search agent in the next iteration. The above iterative process is repeated until the maximum number of iterations is reached. Then, the final optimal solution X * The output is an optimal data locality scheduling plan, which also calculates the optimal bandwidth allocation for each data flow. Finally, the integrated computing scheduler coordinates the computing controller and the network controller to make scheduling decisions based on the scheduling strategy, achieving efficient task execution and optimal resource allocation, thereby obtaining the task scheduling strategy and the coordinated flow transmission scheduling strategy.

[0110] In wide-area distributed computing environments, multiple Coflows are often required to transmit simultaneously. These multiple Coflows may face resource competition and mutual interference. Compared to single-Coflow transmission, multi-Coflow transmission not only requires consideration of the transmission needs of a single Coflow, but also the coordination and balance between multiple Coflows. Therefore, multi-Coflow transmission scheduling requires more complex considerations and greater decision-making difficulty. The optimization goal of multi-Coflow transmission scheduling is typically to improve overall data transmission performance, which can be quantified using the average CCT.

[0111] Based on the single Coflow transmission scheduling, the embodiment of the present application proposes a multi-Coflow transmission rescheduling method to handle dynamically arriving Coflows and improve the overall data transmission capability. Figure 6 Schematic diagram of the multi-Coflow transmission rescheduling process. When a new Coflow arrives, the wide-area scheduling method that takes into account data locality and Coflow transmission is first called to calculate the task scheduling and bandwidth allocation of the new Coflow. Subsequently, the bandwidth of all existing data flows, including the flow in this new Coflow, is readjusted to fully utilize the communication link capacity and fairly distribute network resources among multiple Coflows to avoid excessive resource occupation by some Coflows affecting the transmission of other Coflows, thereby improving the overall data transmission capacity. The specific steps of the multi-Coflow transmission rescheduling method are as follows:

[0112] Step 1: When receiving feedback that a new Coflow has arrived or an existing Coflow has been completed, add the new Coflow to the Coflow queue W or delete the completed Coflow.

[0113] Step 2: In order to avoid frequent computing task scheduling, the wide-area scheduling method that considers data locality and Coflow transmission is called only for the newly arrived Coflow to calculate the task scheduling and bandwidth allocation of the new Coflow, and the optimal solution of data locality scheduling and bandwidth allocation for each Coflow is stored.

[0114] Step 3: Calculate the weight factor for possible bandwidth competition between multiple Coflows To proportionally readjust the bandwidth allocation of each Coflow, where T v The minimum communication time of the Coflow is used to ensure fairness in concurrent flow transmission, so that large concurrent flows get more bandwidth and small concurrent flows get relatively less bandwidth.

[0115] Step 4: Use the weight factor to scale the bandwidth of all data flows based on the weighted bandwidth sharing strategy to fully utilize the remaining bandwidth.

[0116] In order to demonstrate the technical advantages of the method of considering data locality and Coflow transmission scheduling, the embodiment of the present application provides Figure 7 The multi-Coflow transmission scheduling schematic diagram shown in FIG is used to specifically illustrate how the method performs joint optimization of computing task scheduling and Coflow transmission scheduling considering data locality. Figure 7Figure (a) has two coflows. For coflow a, there are two tasks Ta1 and Ta2, which depend on data A1 and A2. For coflow b, there is only one task Tb1, which depends on data B1 and B2. The bandwidth capacity of the link L1 from Data Center 1 to Data Center 2 is 10 Mbps, and the bandwidth capacity of the link L2 from Data Center 2 to Data Center 1 is 2 Mbps. Figure 7 (b) in Figure 1 provides an example of an optimal scheduling plan, which schedules tasks Tb1 and Ta2 to data center 2 and task Ta1 to data center 1. This scheduling plan jointly optimizes computing task scheduling and Coflow transmission scheduling with consideration of data locality, allowing Coflow B to cleverly avoid the bottleneck link L2 with smaller bandwidth. The data flow size of Coflow B is reduced to a minimum of 100 Mb, and the bandwidth resource competition between data flows F2 and F3 generated by Coflow a and Coflow b on link L1 can be resolved through a multi-Coflow transmission rescheduling method, thereby maximizing the overall data transmission performance.

[0117] From the above embodiments, it can be seen that the embodiments of the present application realize the deep integration of computing task scheduling and collaborative data flow scheduling considering data locality through the integrated scheduling method. This integration can give full play to the collaborative optimization effect of multi-dimensional resources. Reduce the processing time of big data tasks from the three aspects of optimizing computing task allocation, data locality scheduling, and collaborative data flow network transmission, improve resource utilization, and achieve efficient execution of tasks and optimal allocation of resources. In addition, the wide-area scheduling device and method considering data locality and Coflow transmission proposed in the embodiments of the present application adopt an online scheduling method based on a meta-heuristic algorithm to perform joint optimization of computing task scheduling and collaborative data flow scheduling considering data locality. This method can provide technical advantages of high efficiency, good scalability and dynamic adaptability in dealing with complex and changeable distributed environments, large-scale networks, and multi-objective optimization. In particular, it can dynamically adjust the scheduling strategy according to real-time changing tasks and resource conditions to ensure that the system always maintains high performance and efficient resource utilization under different operating conditions. In addition, the wide-area scheduling device and method considering data locality and Coflow transmission proposed in the embodiments of the present application can dynamically make Coflow scheduling decisions by monitoring information such as resource status, network topology, traffic characteristics, and task characteristics in real time.

[0118] The above describes various methods of the embodiments of the present application. The following further provides apparatuses for implementing the above methods.

[0119] Please refer to Figure 1 , an embodiment of the present application provides a wide-area scheduling device for task requests, comprising:

[0120] A task parsing module is configured to receive a data processing task request; parse the task request to obtain at least one task corresponding to the data request, attribute data of each task, dependencies between tasks, and requirements of each task; the attribute data may include at least one of the following: task type, task scale, task priority, timeliness requirement, and data source;

[0121] Data management module, used to obtain a global view of data resources;

[0122] Resource management module, used to obtain a global view of physical resources;

[0123] A general computing integrated scheduler is used to formulate a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that considers data locality based on the dependencies between the tasks and the requirements and resource status data of each task; wherein the resource status data includes a global data resource view and a global physical resource view; the computing task scheduling strategy includes the execution nodes of each task, and the single collaborative flow transmission scheduling strategy includes the initial bandwidth of the data flow between nodes and the minimum communication time of the collaborative flow; based on the initial bandwidth and minimum communication time of all collaborative flows, transmission rescheduling optimization of multiple collaborative flows is performed, wherein all collaborative flows include the currently executing collaborative flow and the collaborative flow corresponding to the task request.

[0124] Optionally, the integrated computing scheduler is further configured to:

[0125] Determining a collaborative flow according to the dependency relationship between the tasks, where the collaborative flow includes at least one data flow;

[0126] According to the requirements of each task and the resource status data, a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy considering data locality are determined using a preset algorithm.

[0127] Optionally, the integrated computing scheduler is further configured to:

[0128] Determining a weight factor of each collaborative flow according to the minimum communication time of all collaborative flows;

[0129] Using the weight factor to weight the initial bandwidth of the data flow of each collaborative flow to obtain the updated bandwidth of each data flow; wherein the weight factor is positively correlated with the minimum communication time;

[0130] According to the updated bandwidth of each data flow, transmission scheduling of multiple coordinated flows is performed.

[0131] Optionally, the preset algorithm is a meta-heuristic algorithm.

[0132] Optionally, a preset algorithm is used to determine a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that considers data locality, including:

[0133] Establishing a model for an optimization problem, the model including an objective function and constraints, wherein the objective function is the communication time of a data flow from node i to node j in the collaborative flow, and the communication time is the ratio of the data volume of the data flow to the bandwidth of the data flow; the constraints include at least one of a first constraint, a second constraint, and a third constraint, wherein the first constraint is a constraint on the data volume of the data flow transmitted from node i to node j, the second constraint is a constraint that the bandwidth allocated to the data flow from node i to node j is less than the available bandwidth from node i to node j, and the third constraint is a constraint that the same data partition can only be allocated to one node;

[0134] The model of the optimization problem is solved by using a meta-heuristic algorithm to obtain the computing task scheduling strategy and the single collaborative flow transmission scheduling strategy.

[0135] Optionally, the integrated computing scheduler is further configured to:

[0136] After the task request is processed, the access popularity of the data used in this task request is collected;

[0137] The layout of the data used in the task request on each node is adjusted according to the popularity of the data, the computing power and / or the load of each node.

[0138] It should be noted that the device in this embodiment is a device corresponding to the above-mentioned method, and the implementation methods in the above-mentioned embodiments are all applicable to the embodiments of the device and can achieve the same technical effects. The above-mentioned device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned method embodiment and can achieve the same technical effects. The parts and beneficial effects of this embodiment that are the same as those in the method embodiment will not be specifically described here.

[0139] Another embodiment of the present application is a wide area scheduling system for task requests, such as Figure 8 As shown, it includes a transceiver 1210, a processor 1200, a memory 1220, and a program or instruction stored in the memory 1220 and executable on the processor 1200; when the processor 1200 executes the program or instruction, each process of the above-mentioned method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0140] The transceiver 1210 is configured to receive and send data under the control of the processor 1200 .

[0141] Among them, Figure 8 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically various circuits of one or more processors represented by processor 1200 and memory represented by memory 1220. The bus architecture may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. The bus interface provides an interface. The transceiver 1210 may be a plurality of elements, i.e., a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium. The processor 1200 is responsible for managing the bus architecture and general processing, and the memory 1220 may store data used by the processor 1200 when performing operations.

[0142] The present application also provides a computer-readable storage medium that stores a computer program. When the computer program is executed by a processor, the various processes of the wide-area scheduling method embodiment for the task request are implemented, and the same technical effects are achieved. To avoid repetition, the details are not described here. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0143] An embodiment of the present application also provides a computer program product, including computer instructions. When the computer instructions are executed by a processor, the various processes of the wide-area scheduling method embodiment of the above-mentioned task request are implemented, and the same technical effect can be achieved. To avoid repetition, they will not be repeated here.

[0144] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0145] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, and the computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), including a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, air conditioner, or network equipment, etc.) to execute the methods described in each embodiment of the present application.

[0146] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also make many forms without departing from the purpose of this application and the scope of protection of the claims, all of which are within the protection of this application.

Claims

1. A wide-area scheduling method for task requests, characterized in that: include: Receive data processing task requests; Parsing the task request to obtain inter-task dependencies and requirements of each task of at least one task corresponding to the task request; Acquiring resource status data, the resource status data including a global data resource view and a global physical resource view; Formulate a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that consider data locality based on the dependencies between the tasks and the requirements and resource status data of each task; wherein the computing task scheduling strategy includes the execution nodes of each task, and the single collaborative flow transmission scheduling strategy includes the initial bandwidth of the data flow between the nodes and the minimum communication time of the collaborative flow; Transmission rescheduling optimization of multiple collaborative flows is performed according to initial bandwidths and minimum communication times of all collaborative flows, wherein all collaborative flows include the currently executing collaborative flow and the collaborative flow corresponding to the task request.

2. The method according to claim 1, characterized in that Based on the dependencies between the tasks and the requirements and resource status data of each task, a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that considers data locality are formulated, including: Determining a collaborative flow according to the dependency relationship between the tasks, where the collaborative flow includes at least one data flow; According to the requirements of each task and the resource status data, a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy considering data locality are determined using a preset algorithm.

3. The method according to claim 2, characterized in that Perform transmission rescheduling optimization for multiple co-flows based on the initial bandwidth and minimum communication time of all co-flows, including: Determining a weight factor of each collaborative flow according to the minimum communication time of all collaborative flows; Using the weight factor to weight the initial bandwidth of the data flow of each collaborative flow to obtain the updated bandwidth of each data flow; wherein the weight factor is positively correlated with the minimum communication time; According to the updated bandwidth of each data flow, transmission scheduling of multiple coordinated flows is performed.

4. The method according to claim 2, characterized in that The preset algorithm is a meta-heuristic algorithm.

5. The method according to claim 4, characterized in that Using a preset algorithm, determine the computing task scheduling strategy and the single collaborative flow transmission scheduling strategy that considers data locality, including: Establishing a model for an optimization problem, the model including an objective function and constraints, wherein the objective function is the communication time of a data flow from node i to node j in the collaborative flow, and the communication time is the ratio of the data volume of the data flow to the bandwidth of the data flow; the constraints include at least one of a first constraint, a second constraint, and a third constraint, wherein the first constraint is a constraint on the data volume of the data flow transmitted from node i to node j, the second constraint is a constraint that the bandwidth allocated to the data flow from node i to node j is less than the available bandwidth from node i to node j, and the third constraint is a constraint that the same data partition can only be allocated to one node; The model of the optimization problem is solved by using a meta-heuristic algorithm to obtain the computing task scheduling strategy and the single collaborative flow transmission scheduling strategy.

6. The method according to claim 1, characterized in that Also includes: After the task request is processed, the access popularity of the data used in this task request is collected; The layout of the data used in the task request on each node is adjusted according to the popularity of the data, the computing power and / or the load of each node.

7. A wide-area scheduling device for task requests, characterized in that: include: Task parsing module, used to receive task requests for data processing; Parsing the task request to obtain at least one task corresponding to the data request, attribute data of each task, dependencies between tasks, and requirements of each task; the attribute data includes at least one of the following: task type, task scale, task priority, timeliness requirement, and data source; Data management module, used to obtain a global view of data resources; Resource management module, used to obtain a global view of physical resources; A general computing integrated scheduler is used to formulate a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy that considers data locality based on the dependencies between the tasks and the requirements and resource status data of each task; wherein the resource status data includes a global data resource view and a global physical resource view; the computing task scheduling strategy includes the execution nodes of each task, and the single collaborative flow transmission scheduling strategy includes the initial bandwidth of the data flow between nodes and the minimum communication time of the collaborative flow; based on the initial bandwidth and minimum communication time of all collaborative flows, transmission rescheduling optimization of multiple collaborative flows is performed, wherein all collaborative flows include the currently executing collaborative flow and the collaborative flow corresponding to the task request.

8. The device according to claim 7, characterized in that The integrated computing scheduler is further used to: Determining a collaborative flow according to the dependency relationship between the tasks, where the collaborative flow includes at least one data flow; According to the requirements of each task and the resource status data, a computing task scheduling strategy and a single collaborative flow transmission scheduling strategy considering data locality are determined using a preset algorithm.

9. The device according to claim 8, characterized in that The integrated computing scheduler is further used to: Determining a weight factor of each collaborative flow according to the minimum communication time of all collaborative flows; Using the weight factor to weight the initial bandwidth of the data flow of each collaborative flow to obtain the updated bandwidth of each data flow; wherein the weight factor is positively correlated with the minimum communication time; According to the updated bandwidth of each data flow, transmission scheduling of multiple coordinated flows is performed.

10. The device according to claim 9, characterized in that The preset algorithm is a meta-heuristic algorithm.

11. The device according to claim 10, characterized in that Using a preset algorithm, determine the computing task scheduling strategy and the single collaborative flow transmission scheduling strategy that considers data locality, including: Establishing a model for an optimization problem, the model including an objective function and constraints, wherein the objective function is the communication time of a data flow from node i to node j in the collaborative flow, and the communication time is the ratio of the data volume of the data flow to the bandwidth of the data flow; the constraints include at least one of a first constraint, a second constraint, and a third constraint, wherein the first constraint is a constraint on the data volume of the data flow transmitted from node i to node j, the second constraint is a constraint that the bandwidth allocated to the data flow from node i to node j is less than the available bandwidth from node i to node j, and the third constraint is a constraint that the same data partition can only be allocated to one node; The model of the optimization problem is solved by using a meta-heuristic algorithm to obtain the computing task scheduling strategy and the single collaborative flow transmission scheduling strategy.

12. The device according to claim 7, characterized in that The integrated computing scheduler is further used to: After the task request is processed, the access popularity of the data used in this task request is collected; The layout of the data used in the task request on each node is adjusted according to the popularity of the data, the computing power and / or the load of each node.

13. A wide area scheduling system for task requests, characterized in that: include: A transceiver, a processor, a memory, and a program or instruction stored on the memory and executable on the processor; When the processor executes the program or instruction, the steps of the method according to any one of claims 1 to 6 are implemented.

14. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 6.

15. A computer program product, characterized in that The method comprises computer instructions, which, when executed by a processor, implement the steps of the method according to any one of claims 1 to 6.

Images