Method for safely scheduling computing power resources in computing power network

Abstract

The invention relates to the technical field of computing power scheduling, in particular to a computing power resource security scheduling method in a computing power network, which comprises the following steps: creating a computing power network topology, receiving a computing power task request in the computing power network in real time, identifying the attribute of the computing power task request, and analyzing the emergency degree of the computing power task request based on the attribute of the computing power task request. According to the method, through construction of the computing power network topology and analysis of the computing power task request, when the computing power task request does not accord with the available computing power in the generated node network, scheduling of the computing power of the father node or the child node of the node where the computing power task request is located is configured for the computing power task request; reliable execution of the computing power task request is effectively guaranteed, meanwhile, in the computing power task request processing stage, computing power tasks can be processed more stably and orderly through computing power task request queue generation and computing power network health state analysis, and then long-term stability of the computing power network is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of computing power scheduling, and in particular to a method for securely scheduling computing power resources in a computing power network. Background Art

[0002] With the launch and development of the East-West Computing Project, my country has begun to build computing power hub nodes and national data center clusters on a large scale. This has the following two meanings: on the one hand, it expands the overall computing power scale of the country, and at the same time coordinates regional development through national integrated planning and layout. On the other hand, deploying data centers in the west where there is more green energy is conducive to promoting green, low-carbon and circular development.

[0003] The computing network has provided a strong boost to the application and development of East West Computing. The computing network is a new type of network that integrates and links computing resources including those of various data centers to provide users with ubiquitous computing power to meet their computing needs. However, in the process of development and application, it still faces the important challenges of how to jointly dispatch computing resources of many data centers to improve the utilization rate of computing resources; at the same time, it also faces the important challenges of considering energy consumption costs in the process of computing resource dispatch and ensuring the delay requirements of computing tasks and data security.

[0004] At present, most existing methods only consider the model in which the entire network has only a single node with super computing power resources, such as a data center. However, this model has unavailability problems such as single point failure and difficulty in cross-domain use, and does not fully realize the goal of the computing network to combine the computing power resources of various data centers in various domains. In addition, the existing methods lack attention to security in the computing power scheduling process. During the transmission of computing power scheduling data through the computing power network, the data may be maliciously tampered with and there is a risk of privacy leakage. Summary of the invention

[0005] In view of the above-mentioned shortcomings of the prior art, the present invention provides a method for securely scheduling computing resources in a computing network, which solves the technical problems raised in the above-mentioned background technology.

[0006] The technical solution of the present invention to solve the above technical problems is as follows: A method for securely scheduling computing resources in a computing network, comprising: Create a computing power network topology, receive computing power task requests in the computing power network in real time, identify computing power task request attributes, and analyze the urgency of the computing power task request based on the computing power task request attributes; obtain the urgency analysis results of the computing power task request, generate a computing power task request processing queue in descending order based on the urgency analysis results of the computing power task request, select computing power task requests in the computing power task request processing queue in sequence, and provide computing power resource scheduling services for the selected computing power task requests; traverse the computing power network topology, identify the computing power resources of each node in the computing power network topology, take all nodes in the computing power network topology as parent nodes, accumulate the computing power resources of their child nodes to build a computing power resource pool; identify the selected computing power The node in the computing network topology where the task is located determines the parent node and child node to which the node belongs, and analyzes whether the current remaining computing power resources of the node where the computing task request is located meet the execution of the computing task request; if the analysis result is yes, end and jump to the selection stage of the computing task request; if the analysis result is no, analyze the suitability of the parent node and child node to which the node belongs, and select the node with high suitability as the computing power scheduling target; when the computing power scheduling target is unique, the current remaining computing power in the computing power data pool corresponding to the computing power scheduling target is applied to serve the computing power task request currently being processed; when the computing power scheduling target is not unique, the current remaining computing power in each computing power data pool corresponding to each computing power scheduling target is applied to serve the computing power task request currently being processed.

[0007] Furthermore, during the creation phase of the computing network topology, the location information of each node in the computing network and the correlation between each node are synchronously uploaded. The correlation between each node in the computing network is represented as: child nodes and parent nodes connected to each other. The computing network topology is created based on the location information of each node in the computing network and the correlation between each node. The computing task request in the computing network comes from the network users distributed at the locations of each node in the computing network topology. The computing task request attributes include: CPU demand, memory demand, storage demand, task deadline, and expected execution time. The CPU demand is the number of instructions executed per second by the computing network. Among them, the task deadline and expected execution time in the computing task request attributes are determined based on the network type used by the user who obtains the computing task.

[0008] Furthermore, the urgency of the computing task request is calculated by the following formula: ; Where: The urgency of the computing task request; , , The CPU requirements, memory requirements, and storage requirements requested for computing tasks; , , The maximum available CPU, available memory, and available storage of the network used by the network user that requested the computing task; The task deadline of the network used by the user of the network from which the computing task request comes; The start time of the computing task request; The expected execution time of the network used by the source network user of the computing task request; Among them, the urgency of the computing task request The larger the value, the more urgent the computing task request is; conversely, the smaller the value, the less urgent the computing task request is.

[0009] Furthermore, the task deadline of the network used by the network user from which the computing task request comes is , the expected execution time of the network used by the user of the computing task request source network for: ; Where: The size requested for the computing task; The transmission rate of the network used by the source network user of the current computing task request; The maximum transmission rate of the network used by the source network user of the computing task request; is a constant; Among them, the size of the computing task request Customized by the user, or , the constant value obeys .

[0010] Furthermore, each time a computing power task request is selected in the computing power task request processing queue, a computing power network health status assessment is performed synchronously. When the assessment result is healthy, computing power resource scheduling services are further provided for the selected computing power task request. When the assessment result is negative, the currently selected computing power task request is skipped and the next computing power task request selection operation is executed.

[0011] Furthermore, the computing power network health status evaluation logic is expressed as: ; Where; It is the health performance value of the computing power network; The total amount of computing tasks currently being executed by the computing network; The CPU, memory, and storage used to execute the application for the i-th computing task; The maximum available CPU, memory and storage for the computing network; The number of network failures that occur from the time when the computing network starts to execute computing tasks to the current computing time; The duration from the start of the computing task to the current computing time domain of the computing network; The failure ratio that can be covered by redundant resources of the computing power network; is the total computing network resources; Among them, the computing power network health performance value The larger the value, the healthier the computing network. On the contrary, the healthier the computing network is. The user can customize the computing network health judgment value based on the computing network health judgment value and the computing network health performance value. Compare and determine whether the computing power network is healthy.

[0012] Furthermore, when all end nodes in the computing power network topology act as parent nodes and accumulate the computing power resources of their child nodes to construct a computing power data pool, the constructed computing power resource pool is the computing power resources configured for the end nodes themselves.

[0013] Furthermore, in the stage of analyzing the fitness of the parent nodes and child nodes to which the node belongs, the number of parent nodes to which the node belongs is unique, and the number of child nodes to which the node belongs is one group or more; The analysis logic of the fitness of the node and its parent node, and the fitness of the node and its child node is expressed as: ; Where: for suitability; Request the required computing power for computing tasks; Request the remaining computing power of the node where the computing task is located; It is the collection of parent nodes or child nodes to which the node belongs; is the remaining computing power of the jth node; is the normalization factor; in, The smaller the value, the higher the suitability, and vice versa. The normalization factor >0, select The node with the smallest value is used as the computing power scheduling target, and the analysis of whether the current remaining computing power resources of the node where the computing power task request is located can meet the computing power task execution is performed, that is, Determination of whether it is established.

[0014] Furthermore, after analyzing the fitness of the node and its parent node, and the node and its child node, a set of fitness judgment thresholds are synchronously set, and at least one set of the fitness of the node and its parent node, and the node and its child node meets the fitness judgment threshold. When at least one set meets the fitness judgment threshold, the computing power scheduling operation is further performed. When the fitness of the node and its parent node, and the node and its child node do not meet the fitness judgment threshold, the current computing power task request is processed with the operation performed when the computing power network health status assessment result is no.

[0015] Furthermore, the computing power scheduling target is not unique. When each computing power scheduling target corresponds to each computing power data pool and serves the computing power task request currently being processed, the computing power data pool with a larger current remaining computing power ratio among the computing power data pools will give priority to serving the current computing power task request.

[0016] The beneficial effects of the present invention are: The present invention provides a method for secure scheduling of computing power resources in a computing power network. During the execution of the method, through the construction of the computing power network topology and the analysis of the computing power task request, when the computing power task request does not match the available computing power in the generating node network, the computing power task request is configured with the scheduling of the parent node or child node computing power of the node where the computing power task request is located, thereby effectively ensuring the reliable execution of the computing power task request. At the same time, in the computing power task request processing stage, through the generation of the computing power task request queue and the analysis of the computing power network health status, the computing power tasks can be processed more stably and orderly, thereby ensuring the long-term stability of the computing power network. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The present invention is a flowchart of a method for securely scheduling computing resources in a computing network. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] The present invention will be further described below in conjunction with the embodiments.

[0020] Embodiment 1: A method for securely scheduling computing resources in a computing network in this embodiment is as follows: Figure 1 As shown, including: Create a computing network topology, receive computing task requests in the computing network in real time, identify computing task request attributes, and analyze the urgency of computing task requests based on the computing task request attributes; The urgency of the computing task request is calculated by the following formula: ; Where: The urgency of the computing task request; , , The CPU requirements, memory requirements, and storage requirements requested for computing tasks; , , The maximum available CPU, available memory, and available storage of the network used by the network user that requested the computing task; The task deadline of the network used by the user of the network from which the computing task request comes; The start time of the computing task request; The expected execution time of the network used by the source network user of the computing task request; Among them, the urgency of the computing task request The larger the value, the more urgent the computing task request is; otherwise, the smaller the value, the less urgent the computing task request is. The task deadline of the network used by the user of the computing task request source network , the expected execution time of the network used by the user of the computing task request source network for: ; Where: The size requested for the computing task; The transmission rate of the network used by the source network user of the current computing task request; The maximum transmission rate of the network used by the source network user of the computing task request; is a constant; Among them, the size of the computing task request Customized by the user, or , the constant value obeys ; Obtain the urgency analysis results of the computing task requests, generate a computing task request processing queue in descending order based on the urgency analysis results of the computing task requests, select computing task requests in order from the computing task request processing queue, and provide computing resource scheduling services for the selected computing task requests; Each time a computing task request is selected in the computing task request processing queue, a computing network health status assessment is performed synchronously. When the assessment result is healthy, computing resource scheduling services are further provided for the selected computing task request. When the assessment result is negative, the current selected computing task request is skipped and the selection operation of the next computing task request is performed. The logic of computing network health status evaluation is expressed as: ; Where; It is the health performance value of the computing power network; The total amount of computing tasks currently being executed by the computing network; The CPU, memory, and storage used to execute the application for the i-th computing task; The maximum available CPU, memory and storage for the computing network; The number of network failures that occur from the time when the computing network starts to execute computing tasks to the current computing time; The duration from the start of the computing task to the current computing time domain of the computing network; The failure ratio that can be covered by redundant resources of the computing power network; is the total computing network resources; Among them, the computing power network health performance value The larger the value, the healthier the computing network. On the contrary, the healthier the computing network is. The user can customize the computing network health judgment value based on the computing network health judgment value and the computing network health performance value. Compare and determine whether the computing power network is healthy; Traverse the computing network topology, identify the computing resources of each node in the computing network topology, take all nodes in the computing network topology as parent nodes, accumulate the computing resources of their child nodes to build a computing resource pool; Identify the node in the computing network topology where the selected computing task is located, determine the parent node and child node to which the node belongs, and analyze whether the current remaining computing resources of the node where the computing task request is located can meet the execution of the computing task request; If the analysis result is yes, the process ends and the process jumps to the computing power task request selection stage. If the result of the analysis is no, analyze the suitability of the parent node and child nodes to which the node belongs, and select the node with high suitability as the computing power scheduling target; Analyze the fitness stage of the parent nodes and child nodes to which the node belongs. The number of parent nodes to which the node belongs is unique, and the number of child nodes to which the node belongs is one group or more; The analysis logic of the fitness between a node and its parent node, and between a node and its child nodes is expressed as: ; Where: for suitability; Request the required computing power for computing tasks; Request the remaining computing power of the node where the computing task is located; It is the collection of parent nodes or child nodes to which the node belongs; is the remaining computing power of the jth node; is the normalization factor; in, The smaller the value, the higher the suitability, and vice versa. The normalization factor >0, select The node with the smallest value is used as the computing power scheduling target, and the analysis of whether the current remaining computing power resources of the node where the computing power task request is located can meet the computing power task execution is performed, that is, Determination of whether it is established; After analyzing the fitness of the node and its parent node, and the node and its child node, a set of fitness determination thresholds are set synchronously. At least one set of the fitness of the node and its parent node, and the node and its child node meets the fitness determination threshold. When at least one set meets the fitness determination threshold, the computing power scheduling operation is further performed. When the fitness of the node and its parent node, and the node and its child node do not meet the fitness determination threshold, the current computing power task request is processed with the operation performed when the computing power network health status assessment result is negative; When the computing power scheduling target is unique, the current remaining computing power in the computing power data pool corresponding to the computing power scheduling target is used to serve the computing power task request currently being processed; when the computing power scheduling target is not unique, the current remaining computing power in each computing power data pool corresponding to each computing power scheduling target is used to serve the computing power task request currently being processed; The computing power scheduling target is not unique. When each computing power scheduling target corresponds to each computing power data pool and serves the computing power task request currently being processed, the computing power data pool with the largest current remaining computing power ratio among the computing power data pools will give priority to serving the current computing power task request.

[0021] In this embodiment, through the execution of the method in the above embodiment, an intelligent scheduling service is provided for the computing power resources in the computing power network when serving computing power task requests, ensuring that when computing power task requests are generated in the computing power network, the computing power network can process the computing power task requests more logically and stably, and reasonably manage the computing power resources in the computing power network.

[0022] Embodiment 2: In terms of specific implementation, based on Example 1, this example refers to Figure 1 The following is a further detailed description of a method for safely scheduling computing resources in a computing network in Example 1: During the creation phase of the computing network topology, the location information of each node in the computing network and the correlation between each node are uploaded synchronously. The correlation between each node in the computing network is expressed as: the child nodes and parent nodes connected to each other. The computing network topology is created based on the location information of each node in the computing network and the correlation between each node. The computing task requests in the computing network come from network users distributed at the locations of each node in the computing network topology. The computing task request attributes include: CPU demand, memory demand, storage demand, task deadline, and expected execution time. The CPU demand is the number of instructions executed per second by the computing network. Among them, the task deadline and expected execution time in the computing task request attributes are determined based on the network type used by the user who obtains the computing task.

[0023] Through the above settings, the content of the computing power task request attributes is further limited, and the creation logic of the computing power network topology is further specified.

[0024] like Figure 1 As shown, all end nodes in the computing power network topology act as parent nodes. When accumulating the computing power resources of their child nodes to build a computing power data pool, the constructed computing power resource pool is the computing power resources configured for the end nodes themselves.

[0025] In summary, during the execution of the method in the above embodiment, through the construction of the computing power network topology and the analysis of the computing power task request, when the computing power task request does not match the available computing power in the generating node network, the computing power task request is configured with the scheduling of the parent node or child node computing power of the node where the computing power task request is located, thereby effectively ensuring the reliable execution of the computing power task request. At the same time, in the computing power task request processing stage, through the generation of the computing power task request queue and the analysis of the computing power network health status, the computing power tasks can be processed more stably and orderly, thereby ensuring the long-term stability of the computing power network.

[0026] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for securely scheduling computing resources in a computing network, characterized in that: include: Create a computing network topology, receive computing task requests in the computing network in real time, identify computing task request attributes, and analyze the urgency of computing task requests based on the computing task request attributes; Obtain the urgency analysis results of the computing task requests, generate a computing task request processing queue in descending order based on the urgency analysis results of the computing task requests, select computing task requests in order from the computing task request processing queue, and provide computing resource scheduling services for the selected computing task requests; Traverse the computing network topology, identify the computing resources of each node in the computing network topology, take all nodes in the computing network topology as parent nodes, accumulate the computing resources of their child nodes to build a computing resource pool; Identify the node in the computing network topology where the selected computing task is located, determine the parent node and child node to which the node belongs, and analyze whether the current remaining computing resources of the node where the computing task request is located can meet the execution of the computing task request; If the analysis result is yes, the process ends and the process jumps to the computing power task request selection stage. If the result of the analysis is no, analyze the suitability of the parent node and child nodes to which the node belongs, and select the node with high suitability as the computing power scheduling target; When the computing power scheduling target is unique, the computing power scheduling target corresponds to the current remaining computing power in the computing power data pool to serve the computing power task request currently being processed; When the computing power scheduling target is not unique, each computing power scheduling target is applied to the current remaining computing power in each computing power data pool to serve the computing power task request currently being processed.

2. A method for securely scheduling computing resources in a computing network according to claim 1, characterized in that: During the creation phase of the computing network topology, the location information of each node in the computing network and the correlation between each node are synchronously uploaded. The correlation between each node in the computing network is represented as: child nodes and parent nodes connected to each other. The computing network topology is created based on the location information of each node in the computing network and the correlation between each node. The computing task request in the computing network comes from the network users whose nodes are distributed in the computing network topology. The computing task request attributes include: CPU demand, memory demand, storage demand, task deadline, and expected execution time. The CPU demand is the number of instructions executed per second by the computing network. Among them, the task deadline and expected execution time in the computing task request attributes are determined based on the network type used by the user who obtains the computing task.

3. The method for securely scheduling computing resources in a computing network according to claim 1, characterized in that: The urgency of the computing task request is calculated by the following formula: ； Where: The urgency of the computing task request; , , The CPU requirements, memory requirements, and storage requirements requested for computing tasks; , , The maximum available CPU, available memory, and available storage of the network used by the network user that requested the computing task; The task deadline of the network used by the user of the network from which the computing task request comes; The start time of the computing task request; The expected execution time of the network used by the source network user of the computing task request; Among them, the urgency of the computing task request The larger the value, the more urgent the computing task request is; conversely, the smaller the value, the less urgent the computing task request is.

4. The method for securely scheduling computing resources in a computing network according to claim 3, characterized in that: The task deadline of the network used by the network user from which the computing task request comes , the expected execution time of the network used by the user of the computing task request source network for: ； Where: The size requested for the computing task; The transmission rate of the network used by the source network user of the current computing task request; The maximum transmission rate of the network used by the source network user of the computing task request; is a constant; Among them, the size of the computing task request Customized by the user, or , the constant value obeys .

5. The method for securely scheduling computing resources in a computing network according to claim 1, characterized in that: Each time a computing task request is selected in the computing task request processing queue, a computing network health status assessment is performed synchronously. When the assessment result is healthy, computing resource scheduling services are further provided for the selected computing task request. When the assessment result is negative, the currently selected computing task request is skipped and the next computing task request selection operation is performed.

6. The method for securely scheduling computing resources in a computing network according to claim 1, characterized in that: The computing power network health status evaluation logic is expressed as: ； Where: It is the health performance value of the computing power network; The total amount of computing tasks currently being executed by the computing network; The CPU, memory, and storage used to execute the application for the i-th computing task; The maximum available CPU, memory and storage for the computing network; The number of network failures that occur from the time when the computing network starts to execute computing tasks to the current computing time; The duration from the start of the computing task to the current computing time domain of the computing network; The failure ratio that can be covered by redundant resources of the computing power network; is the total computing network resources; Among them, the computing power network health performance value The larger the value, the healthier the computing network. On the contrary, the healthier the computing network is. The user can customize the computing network health judgment value based on the computing network health judgment value and the computing network health performance value. Compare and determine whether the computing power network is healthy.

7. The method for securely scheduling computing resources in a computing network according to claim 1, characterized in that: All end nodes in the computing power network topology act as parent nodes. When accumulating the computing power resources of their child nodes to build a computing power data pool, the constructed computing power resource pool is the computing power resources configured for the end nodes themselves.

8. The method for securely scheduling computing resources in a computing network according to claim 1, characterized in that: Analyze the fitness stage of the parent nodes and child nodes to which the node belongs. The number of parent nodes to which the node belongs is unique, and the number of child nodes to which the node belongs is one group or more; The analysis logic of the fitness of the node and its parent node, and the fitness of the node and its child node is expressed as: ； Where: for suitability; Request the required computing power for computing tasks; Request the remaining computing power of the node where the computing task is located; It is the collection of parent nodes or child nodes to which the node belongs; is the remaining computing power of the jth node; is the normalization factor; in, The smaller the value, the higher the suitability, and vice versa. The normalization factor >0, select The node with the smallest value is used as the computing power scheduling target, and the analysis of whether the current remaining computing power resources of the node where the computing power task request is located can meet the computing power task execution is performed, that is, Determination of whether it is established.

9. A method for securely scheduling computing resources in a computing network according to claim 8, characterized in that: After analyzing the fitness of the node and its parent node, and the node and its child node, a set of fitness determination thresholds are synchronously set. Among the fitness of the node and its parent node, and the node and its child node, at least one set meets the fitness determination threshold. When at least one set meets the fitness determination threshold, the computing power scheduling operation is further performed. When the fitness of the node and its parent node, and the node and its child node do not meet the fitness determination threshold, the current computing power task request is processed with the operation performed when the computing power network health status assessment result is no.

10. The method for securely scheduling computing resources in a computing network according to claim 1, characterized in that: The computing power scheduling target is not unique. When each computing power scheduling target corresponds to each computing power data pool and serves the computing power task request currently being processed, the computing power data pool with the largest current remaining computing power ratio among the computing power data pools will give priority to serving the current computing power task request.

Images