Computing ad hoc network and scheduling system thereof

Abstract

The invention provides a computing power ad hoc network and a scheduling system thereof, the scheduling system of the computing power ad hoc network is deployed on each computing power node in the computing power ad hoc network in a distributed mode, and the scheduling system comprises a scheduling module, a working module, a database and a mirror image warehouse; the scheduling module is used for task awareness and arrangement scheduling, the working module is used for network awareness, computing power awareness, mirror image management and routing forwarding, the database is used for storing topological relations, computing power information, link performance, task information and a computing power routing table, and the mirror image warehouse is used for storing task mirror image resources. A distributed and decentralized scheduling scheme is adopted, so that not only can the dynamic environment of the computing power ad hoc network be better adapted, but also better fault tolerance and flexibility are achieved. According to the computing power ad hoc network, each computing power node can make a decision autonomously, the computing power nodes can work cooperatively, scattered computing resources can be effectively utilized, the robustness of the computing power ad hoc network is improved, and therefore the complex and changeable computing power requirements are better met.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of self-organizing networks, and specifically to a computing power self-organizing network and a scheduling system thereof. Background Art

[0002] In emergency situations such as war, disaster relief, and emergency response, computing resources are often scarce due to the lack of stable and long-term infrastructure support from operators or service providers. Therefore, it is often necessary to build ad hoc networks using limited computing equipment to provide communication and service capabilities within a small area. In these scenarios, ad hoc computing networks have emerged. In ad hoc computing networks, high node mobility, frequent link disconnections, highly dynamic topologies, limited resources on individual nodes, and distributed computing power are common. Therefore, how to enable nodes in ad hoc computing networks to collaborate, effectively utilize limited computing resources, improve task execution efficiency, and adapt to dynamic network changes has become a pressing issue.

[0003] Currently, centralized scheduling solutions are commonly used, where a centralized scheduling management node is responsible for coordinating and allocating computing tasks to individual computing nodes in the network. However, in the dynamic environment of ad hoc computing networks, changes in network topology increase the management complexity of centralized scheduling. Furthermore, single-point failures in the central scheduling management node can have serious impacts on the entire system, such as reduced task execution efficiency or even paralyzing the entire ad hoc computing network. Consequently, centralized scheduling solutions suffer from poor stability and are unable to meet the complex and ever-changing scheduling requirements of ad hoc computing networks. Summary of the Invention

[0004] The embodiment of the present invention provides a computing power self-organizing network and a scheduling system thereof, so as to solve the problem of poor stability of existing centralized scheduling solutions.

[0005] In a first aspect, an embodiment of the present invention provides a scheduling system for a computing power ad hoc network. The scheduling system is deployed on each computing power node in the computing power ad hoc network in a distributed manner. The scheduling system includes a scheduling module, a working module, a database, and an image repository.

[0006] The scheduling module is used for task awareness and orchestration scheduling. The work module is used for network awareness, computing power awareness, image management, and routing forwarding. The database is used to store topology relationships, computing power information, link performance, task information, and computing power routing tables. The image warehouse is used to store task image resources.

[0007] In response to a task request initiated by a user, the scheduling module of the scheduling node processes the task request, obtains the task image corresponding to the task request from the image repository of the scheduling node, generates task information corresponding to the task request, and stores the task information corresponding to the task request in the database of the scheduling node. The scheduling node is any node in the computing power ad hoc network;

[0008] All computing nodes in the computing power ad hoc network perform network perception and computing power perception through their respective working modules, obtain their corresponding topological relationships, link performance and computing power information, store the obtained information in the database of each computing power node, and synchronize the information obtained by each computing power node to the scheduling node, so that the scheduling node can obtain the global resource information of the computing power ad hoc network;

[0009] The scheduling module of the scheduling node determines the computing power route corresponding to the task request based on the global resource information and the task information corresponding to the task request, and stores it in the computing power routing table in the scheduling node database. At the same time, the computing power route corresponding to the task request is synchronized to the computing power routing table in the database of each computing power node in the computing power ad hoc network;

[0010] Each computing node in the computing power ad hoc network executes tasks according to the computing power routing corresponding to the task request.

[0011] In one embodiment, the global uniqueness of scheduling is ensured by locking. Before the scheduling module processes the task request, it also includes: determining whether the scheduling module has obtained the lock. If not, it waits for the lock to be released in a spin cycle until the lock is obtained before processing the task request.

[0012] In one embodiment, if the scheduling module of the current scheduling node fails, the lock is transferred to the scheduling module of the nearest free node, and the task request is forwarded to the scheduling module of the new node that obtains the lock, and the scheduling module of the new node processes the task request.

[0013] In one embodiment, each computing power node executes a task according to the computing power routing corresponding to the task request, including:

[0014] The working module of each computing power node reads the computing power routing information corresponding to the task request from its own database, and pulls the task image corresponding to the task request from its own image repository based on the obtained computing power routing information;

[0015] The data of the task image is forwarded according to the computing power routing information corresponding to the task request.

[0016] In one embodiment, when the computing power self-organizing network undergoes dynamic changes, each computing power node re-perceives the network and computing power, and obtains new topological relationships, link performance, and computing power information; the scheduling node re-determines the new computing power route corresponding to the task request based on the new global resource information.

[0017] In one embodiment, the dynamic changes of the computing power self-organizing network include one or more of node downtime, changes in the number of nodes, changes in node positions, disconnection of links between nodes, and overflow of node tasks.

[0018] In one embodiment, the computing power information includes processor performance, processor usage, memory size, memory utilization, disk capacity, and disk available space.

[0019] In one embodiment, the topological relationship is used to record the network connection relationship between each computing power node; the link performance includes recording the network performance parameters between the connections between each computing power node, and the network performance parameters include bandwidth, delay and packet loss rate.

[0020] In one embodiment, the scheduling system relies on the operating system of the deployed computing nodes to maintain processes, and the scheduling system communicates based on the networking links of the computing self-organizing network.

[0021] In a second aspect, an embodiment of the present invention provides a computing power self-organizing network, comprising: multiple computing power nodes, the multiple computing power nodes are networked according to a self-organizing network protocol, and each of the multiple computing power nodes is deployed with a scheduling system as described in any one of the first aspects.

[0022] The computing power self-organizing network and its scheduling system provided by the embodiment of the present invention, the scheduling system of the computing power self-organizing network is deployed in a distributed manner on each computing power node in the computing power self-organizing network, and the scheduling system includes a scheduling module, a working module, a database and a mirror warehouse; the scheduling module is used to perform task perception and orchestration scheduling, the working module is used to perform network perception, computing power perception, mirror management and routing forwarding, the database is used to store topology relationships, computing power information, link performance, task information and computing power routing tables, and the mirror warehouse is used to store task mirror resources. In response to a task request initiated by a user, the scheduling module of the scheduling node processes the task request, obtains the task image corresponding to the task request from the image repository of the scheduling node, generates task information corresponding to the task request, and stores the task information corresponding to the task request in the database of the scheduling node, where the scheduling node is any node in the computing power ad hoc network; all computing power nodes in the computing power ad hoc network perform network perception and computing power perception through their respective working modules, obtain their respective topological relationships, link performance and computing power information, store the obtained information in the database of each computing power node, and synchronize the information obtained by each computing power node to the scheduling node so that the scheduling node obtains the global resource information of the computing power ad hoc network; the scheduling module of the scheduling node determines the computing power route corresponding to the task request based on the global resource information and the task information corresponding to the task request, and stores it in the computing power routing table in the scheduling node database, and synchronizes the computing power route corresponding to the task request to the computing power routing table in the database of each computing power node in the computing power ad hoc network; each computing power node in the computing power ad hoc network executes the task according to the computing power route corresponding to the task request. The adoption of a distributed, decentralized scheduling solution not only better adapts to the dynamic environment of the MANET, but also provides greater fault tolerance and flexibility. Each computing node can make its own decisions, and computing nodes can work together to effectively utilize distributed computing resources, improving the robustness of the MANET and better meeting complex and changing computing power requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0024] Figure 1 A schematic diagram of a computing power ad hoc network provided by one embodiment of the present invention;

[0025] Figure 2 A schematic diagram of a scheduling system for a computing power ad hoc network provided by one embodiment of the present invention;

[0026] Figure 3 A schematic diagram of a scheduling node performing task awareness according to an embodiment of the present invention;

[0027] Figure 4A schematic diagram of network perception and computing power perception provided by an embodiment of the present invention;

[0028] Figure 5 A schematic diagram of a scheduling node determining computing power routing according to an embodiment of the present invention;

[0029] Figure 6 A schematic diagram of a task request transfer provided by an embodiment of the present invention;

[0030] Figure 7 A schematic diagram of adjusting computing power routing according to an embodiment of the present invention.

[0031] The above drawings illustrate specific embodiments of the present invention, which will be described in more detail below. These drawings and the accompanying description are not intended to limit the scope of the present invention in any way, but rather to illustrate the concept of the present invention to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0032] The present invention will be further described in detail below by means of specific embodiments in conjunction with the accompanying drawings. Similar elements in different embodiments are numbered with associated similar elements. In the following embodiments, many detailed descriptions are provided to enable the present application to be better understood. However, those skilled in the art will readily appreciate that some of the features may be omitted in different circumstances, or may be replaced by other elements, materials, or methods. In some cases, some operations related to the present application are not shown or described in the specification. This is to avoid the core portion of the present application being overwhelmed by excessive descriptions, and for those skilled in the art, it is not necessary to describe these related operations in detail. They will fully understand the related operations based on the description in the specification and the general technical knowledge in the art.

[0033] In addition, the features, operations, or characteristics described in the specification may be combined in any appropriate manner to form various embodiments. Furthermore, the steps or actions in the method description may be reordered or adjusted in a manner readily apparent to those skilled in the art. Therefore, the various sequences in the specification and drawings are provided solely for the purpose of clearly describing a particular embodiment and are not intended to be mandatory, unless otherwise specified.

[0034] The serial numbers assigned to components herein, such as "first," "second," etc., are used solely to distinguish the objects being described and do not convey any sequential or technical meaning. References to "connection" and "coupling" herein, unless otherwise specified, include both direct and indirect connections (couplings).

[0035] Figure 1 This is a schematic diagram of a computing power ad hoc network provided by an embodiment of the present invention. Figure 1 As shown, the computing power self-organizing network provided by this embodiment includes: multiple computing power nodes, and multiple computing power nodes are networked according to the self-organizing network protocol. Computing power refers to computing ability, that is, the ability to process and execute computing tasks. The computing power in this embodiment can be provided by computing power devices such as computers, mobile communication devices, etc. The computing power self-organizing network provided by this embodiment is a network form in which multiple computing power nodes are connected through a self-organizing network protocol. It has a certain network topology structure, and each computing power node can automatically organize and manage network connections without central control. In the computing power self-organizing network provided by this embodiment, each computing power node (computing node) can dynamically join or leave the network, and the entire network can adaptively adjust the connection method to adapt to changes in the network topology. As Figure 1 As shown, several computing nodes with computing capabilities are networked according to a certain topological relationship through a self-organizing network protocol. Figure 1 The dashed lines between the various computing nodes represent the networking links of the ad hoc computing network. This embodiment does not limit the specific ad hoc networking protocol employed. Using a distributed approach, each computing node in the ad hoc computing network is deployed with a scheduling system for the ad hoc computing network, as provided in any embodiment of the present invention. The scheduling system for the ad hoc computing network provided by the present invention will be further described in detail below through specific embodiments.

[0036] Figure 2 This diagram illustrates a scheduling system for an ad hoc computing network according to one embodiment of the present invention. The scheduling system for the ad hoc computing network provided in this embodiment is deployed in a distributed manner on each computing node in the network. Each computing node is capable of making scheduling decisions and managing its own computing resources. Users can initiate task requests from any node, and decentralized distributed task scheduling ensures maximum resource utilization while ensuring efficient business processing.

[0037] like Figure 2 As shown, the scheduling system of the computing power self-organizing network provided in this embodiment includes a scheduling module (scheduler), a work module (worker), a database (database) and a mirror warehouse. Among them, the scheduling module is used to perform task perception and orchestration scheduling, the work module is used to perform network perception, computing power perception, mirror management and routing forwarding, the database is used to store topological relationships, computing power information, link performance, task information and computing power routing tables, and the mirror warehouse is used to store task mirror resources. It should be noted that the scheduling system of the computing power self-organizing network provided in this embodiment relies on the operating system of the deployed computing power node to maintain the process, and the scheduling system communicates based on the networking link of the computing power self-organizing network.

[0038] The task perception that the scheduling module is responsible for specifically refers to: accepting task requests initiated by users, analyzing the needs, selecting the required task images from the image repository, and combining them to form a task image set; the orchestration and scheduling that the scheduling module is responsible for specifically refers to: based on information such as tasks, networks, and computing power, using the scheduling algorithm to calculate the offloading plan and forwarding routing table for the task image set in the cluster.

[0039] The network perception that the working module is responsible for specifically refers to: sensing the network status and neighbor relationships between nodes, and detecting network changes in real time; the computing power perception that the working module is responsible for specifically refers to: monitoring the node load and changes in computing power resource capacity; the image management that the working module is responsible for specifically refers to: being responsible for operations such as pulling, running, pausing, and deleting images; the routing forwarding that the working module is responsible for specifically refers to: forwarding data between task images in an orderly manner according to the routing table.

[0040] The topological relationship stored in the database is specifically used to record the network connection relationship between each node; the computing power information stored in the database is specifically used to record the information of node computing resources, such as processor performance, memory size, storage capacity, etc.; the link performance stored in the database is specifically used to record the network performance parameters between node connections, such as bandwidth, delay, packet loss rate, etc.; the task information stored in the database is specifically used to record the type, priority, resource requirements, etc. of user tasks, and record the task mirror set; the computing power routing table stored in the database is specifically used to record the data transmission path and forwarding order of task mirrors between different nodes.

[0041] The image repository is specifically used to store the image resources required for various tasks, which can be transplanted to different computing nodes for execution.

[0042] Each computing node in the computing power self-organizing network has the ability to make decisions and schedule, so users can initiate task requests on any node, and any node can process the task requests initiated by users. It should be noted that the node where the user initiates the task request is the scheduling node. In response to the task request initiated by the user, the scheduling module of the scheduling node processes the task request, obtains the task image corresponding to the task request from the image repository of the scheduling node, generates the task information corresponding to the task request, and stores the task information corresponding to the task request in the database of the scheduling node. The scheduling node is any node in the computing power self-organizing network. Task perception is only performed on the scheduling node. Figure 3 Schematic diagram of task perception performed by a scheduling node according to an embodiment of the present invention. Figure 3As shown, the scheduling node achieves task awareness through the following three steps: 1. Accepting task requests; 2. Querying the image repository; and 3. Storing task information. Specifically, in response to a user-initiated task request, the scheduling node first understands and analyzes the task to be executed, including information such as the task's nature, execution time, and resource requirements. It then queries the image repository for the required task images, assembles them into a task image set, and finally stores the task information.

[0043] All computing nodes in the computing power ad hoc network perform network perception and computing power perception through their respective working modules, obtain their corresponding topological relationships, link performance and computing power information, store the obtained information in the database of each computing power node, and synchronize the information obtained by each computing power node to the scheduling node so that the scheduling node can obtain the global resource information of the computing power ad hoc network. It should be emphasized that network perception and computing power perception need to be performed on all computing power nodes in the computing power ad hoc network. Figure 4 This is a schematic diagram of network perception and computing power perception provided by an embodiment of the present invention. Figure 4 As shown, network perception requires monitoring and analysis of the network status of the current node (neighborhood relationships, link bandwidth and latency). Computing power perception obtains the load status of the node, including CPU usage, memory utilization, disk available space, etc. In an optional implementation, the computing power information may include processor performance, processor usage, memory size, memory utilization, disk capacity and disk available space. In an optional implementation, the topological relationship can be used to record the network connection relationship between each computing power node; link performance can include recording network performance parameters between the connections of each computing power node, and network performance parameters include bandwidth, latency and packet loss rate.

[0044] After all computing nodes in the ad hoc computing network complete network and computing power awareness, each computing node synchronizes its neighbor topology, computing power information, and neighbor link information with the scheduling node, allowing the scheduling node to obtain global computing network resource information. The scheduling node's scheduling module then determines the computing power route corresponding to the task request based on the global resource information and the task information corresponding to the task request, stores it in the computing power routing table in the scheduling node database, and simultaneously synchronizes the computing power route corresponding to the task request to the computing power routing table in the database of each computing node in the ad hoc computing network. Figure 5 Schematic diagram of a scheduling node determining computing power routing according to an embodiment of the present invention. Figure 5As shown, first, each computing power node synchronizes topology, computing power, link and other information to the scheduling node through distributed synchronization. Then, the scheduling module of the scheduling node reads the global computing network information and task information, specifically reads the topology, computing power, link, task and other information, and calculates the computing power route of the task through the scheduling algorithm. Finally, the calculated computing power route is stored in the computing power routing table of the database and the computing power routing table is synchronized to each computing power node in the computing power self-organizing network through distributed synchronization. It should be noted that this embodiment does not limit the specific scheduling algorithm used.

[0045] Through the distributed synchronous computing power ad hoc network, each computing power node can obtain the computing power routing table synchronized with the scheduling node. Based on this computing power routing table, the computing power routing corresponding to the task request can be obtained. Finally, each computing power node in the computing power ad hoc network executes the task according to the computing power routing corresponding to the task request. Each computing power node executes the task according to the computing power routing corresponding to the task request. Specifically, the following may be included: the working module of each computing power node reads the computing power routing information corresponding to the task request from its respective database, and based on the obtained computing power routing information, pulls the task image corresponding to the task request from its respective image repository; and forwards the task image data according to the computing power routing information corresponding to the task request. The working module (worker) in the node reads the computing power routing information and pulls the corresponding image to the node. The worker in the node queries the computing power routing information and forwards the task image data to other nodes to facilitate the next step of task processing.

[0046] The scheduling system of the computing power self-organizing network provided in this embodiment is deployed in a distributed manner on each computing power node in the computing power self-organizing network. The scheduling system specifically includes a scheduling module, a working module, a database and a mirror warehouse; the scheduling module is used to perform task perception and orchestration scheduling, the working module is used to perform network perception, computing power perception, mirror management and routing forwarding, the database is used to store topological relationships, computing power information, link performance, task information and computing power routing tables, and the mirror warehouse is used to store task mirror resources. The use of a distributed and decentralized scheduling scheme can not only better adapt to the dynamic environment of the computing power self-organizing network, but also have better fault tolerance and flexibility. Each computing power node can make autonomous decisions, and the computing power nodes can work together, which can effectively utilize distributed computing resources and improve the robustness of the computing power self-organizing network, thereby better meeting complex and changing computing power requirements.

[0047] On the basis of the above-mentioned embodiment, in order to ensure the correct execution of scheduled tasks and further improve the stability of the computing power self-organizing network, the scheduling system of the computing power self-organizing network provided in this embodiment ensures the global uniqueness of the scheduling by locking. Specifically, before the scheduling module processes the task request, it first determines whether the scheduling module has acquired the lock. If not, it waits for the lock to be released in a spin cycle until the lock is acquired before processing the task request. This embodiment does not limit the specific implementation method of locking. It is understandable that in order to ensure the correct execution of the scheduled task, only one scheduler can be executed by a single thread globally. The global uniqueness is maintained by locking, concurrency conflicts are prevented, and it is ensured that only one scheduler thread can perform task scheduling operations at any time. When the scheduler thread wants to execute task scheduling, it first needs to acquire the lock. If the lock is already occupied by another scheduler thread, the task processing of the current scheduler thread will be blocked. This ensures that only one scheduler thread can process the task request at the same time, preventing concurrency conflicts and causing scheduling errors. If the current scheduler thread cannot acquire the lock, it will spin in a loop waiting for the lock to be released until it acquires the lock and obtains the right to process the task request.

[0048] On the basis of any of the above embodiments, in order to ensure the high availability of the scheduling system, in the scheduling system of the computing power self-organizing network provided by this embodiment, if the scheduling module of the current scheduling node fails, the lock will be transferred to the scheduling module of the nearest free node, and the task request will be forwarded to the scheduling module of the new node that obtains the lock, and the scheduling module of the new node will process the task request. Figure 6 Schematic diagram of task request transfer provided by one embodiment of the present invention. Figure 6 As shown in the figure, the scheduling module (scheduler) of the current scheduling node (node ​​3) fails. The scheduling modules (scheduler) of each node preempt the lock on a first-come, first-served basis. The lock is globally unique and is transferred to the scheduling module (scheduler) of node 2. At the same time, the task request is forwarded to the scheduling module (scheduler) of node 2 that has obtained the lock. The scheduling system deployed on node 2 reprocesses the task request, thereby ensuring the high availability of the scheduling system.

[0049] The dynamic changes of the network and computing power will have an impact on the task scheduling system, so the task offloading scheme and traffic routing need to be adjusted accordingly. Based on any of the above embodiments, in order to adapt to the dynamic changes of the network, so as to more effectively utilize limited computing resources and improve the efficiency of task execution, the scheduling system of the computing power self-organizing network provided in this embodiment, when the computing power self-organizing network undergoes dynamic changes, each computing power node re-perceives the network and computing power, obtains new topological relationships, link performance and computing power information; the scheduling node re-determines the new computing power routing corresponding to the task request based on the new global resource information. Figure 7 This is a schematic diagram of adjusting computing power routing according to an embodiment of the present invention. Figure 7 As shown, the computing power of node 3 is insufficient, and task image 2 needs to be migrated from node 3 to node 2. At the same time, the traffic routing also needs to be changed accordingly.

[0050] In an optional implementation, the dynamic changes of the computing power self-organizing network include one or more of node downtime, node quantity change, node position change, inter-node link disconnection and node task overflow. Node downtime refers to the situation where a computing power node may go down due to unexpected circumstances; in addition, the resource utilization rate of tasks on the node may suddenly increase, resulting in the inability to continue normal processing. The change in the number of nodes means that as nodes dynamically join or leave, the self-organizing network has the ability to adapt. For example, when a node leaves the cluster, the connection between related nodes needs to be re-established, and the network topology will change. Since the nodes in the self-organizing network are mobile, the node positions will change, and the network topology needs to be reorganized according to the actual situation. Due to unexpected circumstances, the links between nodes may be disconnected and the networking links may change.

[0051] When one or more of the following conditions occur: node downtime, changes in node number, node location, disconnected links between nodes, or overflow of node tasks, this may result in changes in node available computing power, network topology, or network congestion. At this time, the scheduling node's database will synchronize global computing power, network, and topology information and reschedule. If the computing power routing needs to be modified, the task will be migrated; if not, the task will be executed according to the original computing power routing.

[0052] In summary, this application addresses the complex situations in computing power self-organizing networks, such as strong node mobility, frequent link disconnection, high topology dynamics, limited node resources, and dispersed computing power. It increases the complexity of computing power measurement, computing network perception, and task scheduling, and pays special attention to considering the unstable changes of factors such as network topology and node spatial location, ensuring that resources are utilized to the maximum extent while efficiently processing business, and can dynamically adjust the scheduling strategy and perform task migration according to changes in computing power and network conditions. At the same time, the system enables each node in the cluster to have the ability to schedule through decentralized distributed computing power scheduling, which can effectively avoid problems such as single point failure caused by centralized scheduling. In the computing power self-organizing network environment, the scheduling system is deployed on each computing power node in a distributed manner. By perceiving the node computing power and the cluster network, the decentralized distributed computing power scheduling system is used to process task requests initiated by any node, and the scheduling strategy can be adjusted according to the dynamic changes of computing power and the network. The use of a distributed and decentralized scheduling scheme can better adapt to dynamic environments and have better fault tolerance and flexibility. Nodes can make autonomous decisions and work collaboratively, effectively utilizing distributed computing resources, improving the robustness and adaptability of the system, and thus better meeting complex and changing computing power requirements.

[0053] The various embodiments in the present disclosure are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0054] The scope of protection of the present disclosure is not limited to the above-described embodiments. Obviously, those skilled in the art may make various modifications and variations to the present disclosure without departing from the scope and spirit of the present disclosure. If such modifications and variations fall within the scope of the claims of the present disclosure and their equivalents, the present disclosure is intended to include such modifications and variations.

Claims

1. A scheduling system for a computing power ad hoc network, characterized in that: The scheduling system is deployed on each computing node in the computing power self-organizing network in a distributed manner. The scheduling system includes a scheduling module, a working module, a database and an image warehouse; The scheduling module is used to perform task perception and orchestration scheduling, the working module is used to perform network perception, computing power perception, image management and routing forwarding, the database is used to store topology relationships, computing power information, link performance, task information and computing power routing tables, and the image warehouse is used to store task image resources; In response to a task request initiated by a user, a scheduling module of a scheduling node processes the task request, obtains a task image corresponding to the task request from an image repository of the scheduling node, generates task information corresponding to the task request, and stores the task information corresponding to the task request in a database of the scheduling node, where the scheduling node is any node in the computing power ad hoc network; All computing nodes of the computing power ad hoc network perform network perception and computing power perception through their respective working modules, obtain their corresponding topological relationships, link performance and computing power information, store the obtained information in the database of each computing power node, and synchronize the information obtained by each computing power node to the scheduling node, so that the scheduling node obtains the global resource information of the computing power ad hoc network; The scheduling module of the scheduling node determines the computing power route corresponding to the task request based on the global resource information and the task information corresponding to the task request, and stores it in the computing power routing table in the scheduling node database, and synchronizes the computing power route corresponding to the task request to the computing power routing table in the database of each computing power node in the computing power ad hoc network; Each computing power node of the computing power self-organizing network executes the task according to the computing power routing corresponding to the task request.

2. The dispatching system according to claim 1, characterized in that: The global uniqueness of scheduling is ensured by locking. Before the scheduling module processes the task request, it also includes: judging whether the scheduling module has acquired the lock. If not, it waits for the lock to be released in a spin cycle until the lock is acquired before processing the task request.

3. The dispatching system according to claim 2, characterized in that: If the scheduling module of the current scheduling node fails, the lock is transferred to the scheduling module of the nearest free node, and the task request is forwarded to the scheduling module of the new node that obtains the lock, and the scheduling module of the new node processes the task request.

4. The dispatching system according to claim 1, characterized in that: Each computing power node executes the task according to the computing power routing corresponding to the task request, including: The working module of each computing power node reads the computing power routing information corresponding to the task request from its own database, and pulls the task image corresponding to the task request from its own image repository based on the obtained computing power routing information; The data of the task image is forwarded according to the computing power routing information corresponding to the task request.

5. The dispatching system according to claim 1, characterized in that: When the computing power self-organizing network undergoes dynamic changes, each computing power node re-perceives the network and computing power to obtain new topology relationships, link performance, and computing power information; The scheduling node re-determines a new computing power route corresponding to the task request based on the new global resource information.

6. The dispatching system according to claim 5, characterized in that: The dynamic changes of the computing power self-organizing network include one or more of node downtime, change in the number of nodes, change in node location, disconnection of links between nodes, and overflow of node tasks.

7. The dispatching system according to any one of claims 1 to 6, characterized in that: The computing power information includes processor performance, processor usage, memory size, memory utilization, disk capacity, and disk available space.

8. The dispatching system according to any one of claims 1 to 6, characterized in that: The topological relationship is used to record the network connection relationship between each computing power node; the link performance includes recording the network performance parameters between the connections between each computing power node, and the network performance parameters include bandwidth, delay and packet loss rate.

9. The dispatching system according to any one of claims 1 to 6, characterized in that: The scheduling system relies on the operating system of the deployed computing power node to maintain the process, and the scheduling system communicates based on the networking link of the computing power self-organizing network.

10. A computing power self-organizing network, characterized in that: include: Multiple computing nodes, wherein the multiple computing nodes are networked according to a self-organizing network protocol, and each of the multiple computing nodes is deployed with a scheduling system as described in any one of claims 1 to 9.

Images