A computing power network intelligent scheduling method, system, device and medium

Abstract

This invention provides a method, system, device, and medium for intelligent scheduling of computing power networks, belonging to the field of computing power resource management technology. The intelligent scheduling system's modules work collaboratively. First, it defines access policies and service paths, connecting computing, storage, and algorithm resources to form a computing power resource pool. Then, based on real-time business needs, it uses SRv6 programming capabilities to generate scheduling policies and schedule traffic to the corresponding resource pool. Next, it creates independent computing power network slices according to the policies and maintains transmission, establishing isolation between slices while monitoring data flow performance. In case of anomalies, it initiates intelligent operation and maintenance and outputs optimization suggestions. This invention achieves efficient integration and scheduling of computing power resources, ensuring that resources are allocated to services on demand; it enhances service security and independence through network slicing and isolation mechanisms; real-time monitoring and intelligent operation and maintenance reduce the impact of faults, optimize system performance, and improve the utilization efficiency and service quality of the computing power network.

Description

Technical Field

[0001] This invention belongs to the field of computing power resource management technology, specifically relating to a method, system, device and medium for intelligent scheduling of computing power networks. Background Technology

[0002] In the era of booming digital economy, computing power has risen to become a core productive force, permeating all sectors of society. Computing Force Network (CFN), as a new type of information infrastructure, integrates key elements such as computing power, network, data, and intelligence. Through the integration, perception, control, and management of these elements, it can accurately schedule user resource requests to the optimal resource nodes, significantly improving the overall network resource utilization rate and achieving global scheduling while ensuring the quality of service for resource users.

[0003] In computing networks, industrial control commands require multi-hop routing, resulting in high and unstable path latency. Resource access relies on initial configuration, making it unsuitable for the computing power needs of different production lines. Slicing isolation only supports dedicated physical links, leading to low resource utilization. Video rendering requires high bandwidth, but network slicing only supports fixed bandwidth allocation, failing to handle sudden traffic surges. When multiple tenants share links, communication bandwidth cannot be effectively matched based on priority, causing rendering delays. Data analysis requires multiple tenants to share computing resources, but if the resource pool cannot adjust node allocation, some tenants will have idle resources while others will have insufficient resources, failing to meet user needs. Summary of the Invention

[0004] This invention provides an intelligent scheduling method for computing power networks. The method enables efficient integration and scheduling of computing power resources, ensuring that resources are allocated to services on demand; optimizing system performance, and improving the utilization efficiency and service quality of computing power networks.

[0005] The methods include:

[0006] Step S101: In the intelligent scheduling system for computing power networks, define the access policy and service path of the computing power network, wherein the intelligent scheduling system for computing power networks includes a computing power resource management module, a network resource management module, a business demand perception module, an intelligent scheduling module, and a security protection module;

[0007] Step S102: Connect computing resources, storage resources, and algorithm resources to the intelligent scheduling system of computing network via the network resource management module to form a computing resource pool;

[0008] Step S103: Based on the business traffic and application requirements monitored in real time by the business demand perception module, the intelligent scheduling module generates a business scheduling strategy through the programming capability of SRv6, and schedules the business traffic to the corresponding computing resource pool along the service path.

[0009] Step S104: The network resource management module creates independent computing network slices in the computing network using network slicing technology according to the service scheduling strategy, and maintains data transmission within each slice;

[0010] Step S105: The security protection module establishes an isolation method between the computing power network slices;

[0011] Step S106: Monitor the performance of data streams passing through the computing power network slices in real time. When an anomaly is detected, initiate the intelligent operation and maintenance process to troubleshoot and repair the computing power network, and output performance optimization suggestions.

[0012] Preferably, step S101 specifically includes the following steps:

[0013] Deploy a computing power network intelligent scheduling system module at preset nodes;

[0014] Each node is assigned a unique segment identifier (SID), which contains the node's location information, resource type information, and service level information.

[0015] Obtain the service type, bandwidth requirements, and latency sensitivity, and arrange the allocated segment identifiers into a segment list according to the service requirement path order to form a service path from the service source node to the destination computing node;

[0016] The defined service path is validated by sending probe messages to key nodes in the service path to detect the path's connectivity, latency, and packet loss rate.

[0017] Preferably, step S102 specifically includes the following steps:

[0018] The network resource management module scans the nodes in the computing network using the SNMPv3 protocol or the Netconf protocol to collect resource information from each node.

[0019] The collected resource information is preprocessed; a unique segment identifier (SID) is assigned to each processed resource information, and the node location and resource type of the resource information are configured and encoded into the SID;

[0020] Configure the resource information and corresponding SID into the computing resource management module, and verify the pre-send status of the resource information through the preset destination address of SRv6: send ICMP probe messages to the node where the resource information is located to detect round-trip latency, packet loss rate and bandwidth utilization; after verification, mark the resource information as available and add it to the computing resource pool to complete the resource access process.

[0021] Preferably, step S103 specifically includes the following steps:

[0022] The business demand perception module collects key characteristic parameters of business traffic in real time and encapsulates the parameters into demand tags.

[0023] The intelligent scheduling module calls the interfaces of the computing power resource management module and the network resource management module to obtain the real-time status information of the computing power resource pool, and at the same time extracts the service path corresponding to each node;

[0024] Match business requirement tags with resource status information; the matching methods include: filtering resource information that meets latency sensitivity and bandwidth requirements;

[0025] Based on the node location, a scheduling method is generated that includes the target computing node, transmission path, and bandwidth allocation;

[0026] The feasibility of the scheduling method is verified by sending test messages with service tags to key nodes of the target path to detect the actual latency, packet loss rate and bandwidth availability of the path.

[0027] Preferably, step S104 specifically includes the following steps:

[0028] The network resource management module parses out the service type, resource pool number and service quality level based on the service segment identifier in the service scheduling strategy, and generates network slice information tags accordingly.

[0029] Based on the network slice information tags, the network resource management module configures a virtual bearer channel on the physical network that uniquely corresponds to the service type;

[0030] Within the virtual bearer channel, the network resource management module exclusively reserves link bandwidth, cache depth, and forwarding time slots according to the quality of service level, so that different virtual bearer channels on the same physical link form a zero-preemption relationship;

[0031] When the service scheduling strategy changes, the network resource management module reconfigures the boundaries and reserved resources of the virtual bearer channel in place by configuring incremental fields in the network slice information tag.

[0032] Preferably, step S105 specifically includes the following steps:

[0033] The security protection module assigns a slice-level session key derived from the virtual root key to each computing power network slice;

[0034] The session key K is generated by the virtual root key Kr and the network slice information label D through a one-way function F, where K=F(Kr,D);

[0035] Insert an isolation tag based on the session key at the entry node of each slice;

[0036] When the slice lifecycle ends, the security protection module revokes the corresponding session key and broadcasts a key revocation message;

[0037] During slice operation, the security protection module periodically rotates the session key, and the new key after rotation remains bound to the same network slice information tag.

[0038] Preferably, the method includes: step S106 specifically includes the following steps:

[0039] Pluggable detection probes are set up along the data flow path of the computing power network slice;

[0040] After identifying the diagnostic marker, the node records the passing time in the locally maintained diagnostic timestamp table and associates the diagnostic marker with the node's own hardware counter value to form the raw data of hop-by-hop latency and packet loss indication;

[0041] After receiving a message with raw data, the cloud computing center locates the link segment where the anomaly occurred by comparing the diagnostic timestamps and counter values ​​of adjacent nodes, and generates a compressed repair command containing the anomaly segment identifier and suggested actions. The compressed repair command directly reaches the anomaly segment through the reverse SRv6 path, achieving rapid repair with zero additional routing overhead.

[0042] After the repair is completed, the cloud computing center records the mapping results of anomaly characteristics and repair strategies in the local performance optimization suggestion library.

[0043] This application also provides a computing power network intelligent scheduling system, the system comprising:

[0044] The strategy definition module is used to define the access policy and service path of the computing power network in the intelligent scheduling system of computing power network, wherein the intelligent scheduling system of computing power network includes a computing power resource management module, a network resource management module, a business demand perception module, an intelligent scheduling module and a security protection module;

[0045] Computing resources, storage resources, and algorithm resources are connected to the intelligent scheduling system of the computing power network via the network resource management module to form a computing power resource pool;

[0046] Based on the business traffic and application requirements monitored in real time by the business demand perception module, a business scheduling strategy is generated through the programming capabilities of SRv6, and the business traffic is scheduled to the corresponding computing resource pool along the service path.

[0047] According to the service scheduling strategy, the network resource management module creates independent computing network slices in the computing network through network slicing technology, and maintains data transmission within each slice;

[0048] The security protection module establishes an isolation mechanism between the computing power network slices;

[0049] The system monitors the performance of data streams passing through the computing network slices in real time. When an anomaly is detected, it initiates an intelligent operation and maintenance process to troubleshoot and repair the computing network and outputs performance optimization suggestions.

[0050] According to another embodiment of this application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of a computing power network intelligent scheduling method.

[0051] According to another embodiment of this application, a storage medium is also provided, on which a computer program is stored, wherein the computer program, when executed by a processor, implements the steps of the intelligent scheduling method for computing power networks.

[0052] As can be seen from the above technical solutions, the present invention has the following advantages:

[0053] The intelligent scheduling method for computing power networks provided by this invention achieves on-demand scheduling of computing power networks through SRv6, network slicing, and intelligent operation and maintenance. Specifically, it allocates dedicated paths for control services through SRv6 segment routing, reducing latency. Based on the real-time monitoring of the latency sensitivity of production line instructions by the service awareness module, the intelligent scheduling module adjusts the path; it creates hard isolation slices to ensure the reliability of instruction transmission, and uses extended header isolation identifiers to prevent other services from preempting the transmission; it monitors link latency in real time, and if an anomaly is detected, it immediately switches to a backup path to avoid downtime.

[0054] Based on the business awareness module, bandwidth requirements and latency sensitivity of video rendering are monitored to generate scheduling strategies. Soft-isolation slices are created to reserve bandwidth for high-priority rendering tasks. Bandwidth utilization is monitored in real time; if high-priority tasks lack sufficient bandwidth, bandwidth is borrowed from low-priority slices to ensure smooth rendering. Shared slices are created to achieve multi-tenant logical isolation, and node load rates are monitored in real time, migrating some tasks to redundant nodes within the same slice. This achieves efficient utilization of computing network resources, secure and reliable data transmission, and improves the overall operational efficiency and resilience of the computing network. Attached Figure Description

[0055] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 Here is a flowchart of the intelligent scheduling method for computing power networks;

[0057] Figure 2This is a flowchart illustrating an embodiment of the intelligent scheduling method for computing power networks.

[0058] Figure 3 Flowchart of another embodiment of the intelligent scheduling method for computing power networks;

[0059] Figure 4 This is a schematic diagram of an electronic device. Detailed Implementation

[0060] The intelligent scheduling method for computing power networks disclosed in this application, through the high integration of multiple core modules such as computing power resource management, network resource management, business demand perception, intelligent scheduling strategies, and security protection, aims to achieve efficient, flexible, rapid response, lossless, and secure intelligent scheduling of computing power resources. The system possesses powerful real-time monitoring capabilities, accurately capturing changes in business traffic and the specific needs of various applications. Based on this real-time data, it intelligently adjusts and optimizes the allocation and deployment of computing power resources, ensuring that every business demand is met promptly and fully. Simultaneously, the system implements strict security protection measures throughout the entire process of data transmission and computation processing, effectively guaranteeing the security and integrity of data.

[0061] The intelligent scheduling method for computing power networks involved in this application will be described in detail below. Specific details such as particular system structures and technologies are proposed for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can also be implemented in other embodiments without these specific details.

[0062] It should be understood that, when used in this specification, the term "comprising" indicates the presence of the described feature, integral, step, operation, element, and / or component, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0063] The terms "one embodiment" or "some embodiments" used in this application mean that one or more embodiments of this application include the specific features, structures, or characteristics described in that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this application do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized.

[0064] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0065] Please see Figure 1 The diagram shows a flowchart of a computing power network intelligent scheduling method in a specific embodiment. The method includes:

[0066] Step S101: In the intelligent scheduling system for computing power networks, the access policy and service path of the computing power network are defined through SRv6 technology. The intelligent scheduling system for computing power networks includes a computing power resource management module, a network resource management module, a service demand perception module, an intelligent scheduling module, and a security protection module.

[0067] In some embodiments, within the intelligent scheduling system for computing power networks, the specific functional boundaries of the computing power resource management module, network resource management module, service demand perception module, intelligent scheduling module, and security protection module are first defined. Each module is physically connected via a high-speed fiber optic interface to ensure data transmission rates. Based on SRv6 technology, access strategies are formulated for different types of access devices; for example, a priority access mechanism is used for industrial control equipment, while an access mechanism is used for office equipment. Simultaneously, based on the network topology and service transmission requirements, the node hopping order of the service path is designed. Each path node is configured with a unique SRv6 segment identifier, and at least three backup nodes are set up in the path to handle occasional failures.

[0068] Step S102: Using the segment routing function of the SRv6 technology, computing resources, storage resources and algorithm resources are connected to the intelligent scheduling system of computing network via the network resource management module to form a computing resource pool.

[0069] In some embodiments, the network resource management module first collects information on the server's CPU model, number of cores, hard disk capacity, read / write speed, algorithm application scenarios, and computing efficiency, and establishes a resource information database.

[0070] Utilizing the segment routing function of SRv6 technology, access paths are planned for various resources based on resource type and node location. Each forwarding node in the path guides resource transmission according to the SRv6 segment identifier. During resource access, the access status is monitored in real time, and if access fails, a backup access path is switched to. Successfully accessed resources are categorized and integrated according to computing, storage, and algorithm types to form a computing resource pool, and the resource status within the pool is updated in real time.

[0071] As can be seen, by collecting information to understand resource details, enabling targeted access to resources based on the SRv6 segment routing function, and then classifying and integrating them into a centrally managed resource pool, resources can be scheduled by the system. Multi-path access ensures the stability of resource access and improves resource utilization.

[0072] Step S103: Based on the business traffic and application requirements monitored in real time by the business demand perception module, the intelligent scheduling module generates a business scheduling strategy through the programming capabilities of SRv6, and schedules the business traffic to the corresponding computing resource pool along the service path.

[0073] In this embodiment, the service demand awareness module collects real-time data such as the size and transmission frequency of service traffic through a traffic monitor deployed at the network ingress. Simultaneously, it interfaces with application systems to obtain information such as the minimum bandwidth requirements for video processing services and the maximum latency limits for industrial control services. After receiving this information, the intelligent scheduling module, combined with the real-time status of the computing resource pool, utilizes the programmability of SRv6 to embed information such as service priority and required resource types into the segment list, generating a service scheduling strategy. Following the segment identifier order of the service path, the service traffic is encapsulated into data packets with corresponding identifiers and transmitted to the matching computing resource pool. This embodiment generates an adaptive scheduling strategy based on the programmability of SRv6, ensuring that service traffic is accurately transmitted to the target resource pool along a preset service path, achieving service-resource matching and improving service processing efficiency.

[0074] Step S104: The network resource management module creates independent computing network slices in the computing network using network slicing technology according to the service scheduling strategy, and maintains data transmission within each slice.

[0075] This embodiment determines resource requirements based on the parameters of the service scheduling strategy, divides independent slices and allocates resources through network slicing technology, and configures transmission rules to ensure that data is transmitted in an orderly manner within the slices.

[0076] Step S104 specifically includes the following steps:

[0077] Step S1041: The network resource management module parses out the service type, resource pool number and service quality level based on the service segment identifier in the service scheduling strategy, and generates network slice information tags accordingly. The network slice information tags use the service segment identifier as the key and the service quality level as the value to form the basis for slice creation.

[0078] Step S1042: Based on the network slice information tag, the network resource management module configures a virtual bearer channel on the physical network that uniquely corresponds to the service type. The boundary of the virtual bearer channel is defined by the start node and end node of the service segment identifier.

[0079] Step S1043: Within the virtual bearer channel, the network resource management module exclusively reserves the link bandwidth, cache depth, and forwarding time slots according to the quality of service level, so that different virtual bearer channels on the same physical link form a zero-preemption relationship, thereby realizing lossless transmission within the slice.

[0080] Step S1044: When the service scheduling strategy changes, the network resource management module reconfigures the boundaries and reserved resources of the virtual bearer channel in place by configuring incremental fields in the network slice information tag. This allows for elastic resource adjustment without dismantling the original slice, thus maintaining service continuity.

[0081] It should be noted that the incremental field configuration method allows defining the incremental field format as: Incremental Symbol Information Number, New Start ID, New End ID, New Bandwidth Value. The Incremental Symbol Information Number increments by 1 with each policy change, used to trigger in-situ reconfiguration of the channel.

[0082] It can be seen that the service-level segment identifier carried by the service scheduling policy is directly mapped to network slice information tags by the network resource management module. The information tag uniquely determines the topology range, resource reservation specifications, and service quality level of the virtual bearer channel. Policy changes trigger the information tag version number to increment automatically, and the information tag version number drives the in-situ reconfiguration of the virtual bearer channel. The virtual bearer channel uses SRv6 as the forwarding instruction, and the nodes along the way identify and apply the corresponding reserved resources according to the slice information tag index field in SRv6 to complete lossless transmission within the slice.

[0083] Step S105: The security protection module establishes an isolation method between the computing power network slices.

[0084] In this embodiment, the security protection module first obtains the service security level corresponding to each computing power network slice and determines the isolation strength based on the security level. For high-security-level slices, independent servers and switches are used, and data transmission uses a combination of AES-256 encryption algorithm. For medium- and low-security-level slices, virtual local area networks are used. The module monitors the communication between slices in real time, and immediately blocks and alerts if unauthorized access attempts are detected. This prevents illegal data interaction between slices and ensures the independence and security of each slice.

[0085] Step S105 in this embodiment specifically includes the following steps:

[0086] Step S1051: The security protection module assigns a slice-level session key derived from the virtual root key to each computing power network slice. The session key is irreversibly bound to the network slice information tag of the slice, so that different slices cannot decrypt each other's traffic even if they share the same physical resources.

[0087] Optionally, the session key K is generated by the virtual root key Kr and the network slice information tag D through a one-way function F, K=F(Kr,D), where F is irreversible and the output length is fixed at 256 bits.

[0088] Step S1052: Insert an isolation label generated based on the session key at the ingress node of each slice. The isolation label is embedded in the programmable field of the SRv6 message. Nodes along the way decide whether to forward the message based solely on the matching result of the isolation label. Hard isolation of traffic can be achieved without configuring ACLs on the physical port.

[0089] Step S1053: When the slice lifecycle ends or the policy changes, the security protection module revokes the corresponding session key and broadcasts a key revocation message. After receiving the message, all nodes clear the locally stored session key copy and isolation label mapping to ensure that residual traffic cannot continue to flow.

[0090] The key revocation message in this embodiment includes a revocation list and a revocation sequence number. The revocation list lists all session key identifiers to be revoked, and the revocation sequence number is a monotonically increasing 64-bit integer used to prevent replay attacks.

[0091] Step S1054: During the slice operation, the security protection module periodically rotates the session key. The new key after rotation is still bound to the same network slice information tag, but has no derivation relationship with the previous key, thereby achieving continuous isolation without interrupting services.

[0092] As can be seen, with the session key as the core isolation element, the key itself represents the slice identity by binding the key to the network slice information label in one go. The entry node calculates the isolation label with the session key and writes it into the SRv6 message. Nodes along the way only need to check whether the isolation label matches the local key copy to complete the passage or discard, without the need to maintain a complex access control list. Key revocation and rotation are completed through key revocation messages and periodic update messages.

[0093] Step S106: The intelligent scheduling system for the computing power network monitors the performance of the data stream passing through the computing power network slice in real time. When an anomaly is detected, the intelligent operation and maintenance process is initiated to troubleshoot and repair the computing power network, and performance optimization suggestions are output.

[0094] This embodiment deploys performance monitors at key nodes of the computing power network slice to collect real-time performance metrics such as latency, packet loss rate, and bandwidth utilization of the data stream. The collected data is compared with preset thresholds; when a metric exceeds the threshold, it is considered abnormal. An intelligent operation and maintenance process is then initiated, analyzing the abnormal data to locate the fault point, and then taking corresponding remedial measures. After the repair is completed, performance optimization suggestions are generated and pushed to the operation and maintenance personnel. This embodiment improves the efficiency of fault repair and ensures the long-term stable operation of the system.

[0095] Step S106 specifically includes the following steps:

[0096] Step S1061: Set up pluggable detection probes on the data flow path of the computing power network slice. The detection probes are configured with lightweight diagnostic tags in the packet header that can be recognized by nodes along the way, so that the diagnostic process is transparent to the business and has no additional bandwidth overhead.

[0097] In this embodiment, the diagnostic marker is 32 bits long, with the first 16 bits being the probe identifier and the last 16 bits being the serial number. The serial number increments sequentially according to the message order and is used to uniquely identify a single testing cycle.

[0098] Step S1062: After identifying the diagnostic marker, the node records the passing time in the locally maintained diagnostic timestamp table and associates the diagnostic marker with the node's own hardware counter value to form raw data of hop-by-hop latency and packet loss indication. All raw data is returned to the cloud computing center along with the message.

[0099] Step S1063: After receiving the message with the original data, the cloud computing center locates the link segment where the anomaly occurred by comparing the diagnostic timestamps and counter values ​​of adjacent nodes, and generates a compressed repair instruction containing the anomaly segment identifier and suggested actions. The compressed repair instruction directly reaches the anomaly segment through the reverse SRv6 path, achieving rapid repair with zero additional routing overhead.

[0100] Step S1064: After the repair is completed, the cloud computing center records the mapping results of abnormal features and repair strategies in the local performance optimization suggestion library. When the same feature reappears, the mapping results are called to issue a pre-repair instruction, forming a continuous self-learning performance optimization closed loop.

[0101] This embodiment uses a pluggable detection probe to collect hop-by-hop performance data in real time on the data plane. The collection results are returned along with the service traffic. The cloud computing center uses the returned data to locate abnormal segments and generate compressed repair instructions. The repair instructions reach the problem node along the reverse SRv6 segment route. In this embodiment, the probe uses a fixed-length diagnostic tag embedded in the SRv6 extended header, the node hardware counter provides nanosecond-level timestamps, and the cloud computing center maintains anomaly features and a mapping table to realize anomaly detection.

[0102] In one embodiment of the present invention, based on step S101, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner. For example... Figure 2 As shown, step S101 specifically includes the following steps:

[0103] Step S1011: Deploy the intelligent scheduling system module of computing power network on the preset node, wherein each module realizes direct interconnection between the physical layer and the data link layer through a high-speed communication interface.

[0104] It should be noted that the physical layer and data link layer of the module deployment are interconnected through an Ethernet frame structure. The frame header contains the source module address, destination module address, and frame type field to ensure accurate data transmission between modules and to ensure low latency and high reliability of data transmission between modules.

[0105] Step S1012: Based on SRv6 technology, each node is assigned a unique segment identifier (SID). The segment identifier contains the node's location information, resource type information, and service level information.

[0106] The nodes in this embodiment include a cloud computing center and multiple edge computing nodes, or they can be computing nodes. The allocation method of the segment identifier (SID) can be set as follows: SID = 16-bit binary, computing resources are 01, storage resources are 10, algorithm resources are 11, and deployment level code, etc., to form a 32-bit unique identifier.

[0107] Step S1013: Obtain the service type, bandwidth requirements, and latency sensitivity. Utilize the segment routing function of SRv6 to arrange the allocated segment identifiers into a segment list according to the service requirement path order, forming a service path from the service source node to the destination computing node.

[0108] In this embodiment, the segment list configuration method can be based on the arrangement of the segment list and the network scheduling weights can be configured according to the data resource requirements of the network scheduling. The configuration method of the network scheduling weights is: Zy = α × MG + β × XQ, where MG is the service latency sensitivity, and the higher the sensitivity, the larger the MG value; XQ is the service bandwidth requirement, and the larger the requirement, the larger the XQ value. α and β are the network scheduling adjustment coefficients for data resource requirements, and α + β = 1. The intelligent scheduling module selects the path with the smallest total weight in the segment list as the service path.

[0109] Step S1014: Verify the validity of the defined service path by sending probe messages to key nodes in the service path to detect the connectivity, latency and packet loss rate of the path, and ensure that the service path meets the minimum performance indicators required by the business.

[0110] In step S101 of this embodiment, SRv6 is used to allocate segment identifiers containing node location, resource type and service level to computing nodes; according to the type, bandwidth and latency sensitivity provided by the service requirement perception module, the segment identifiers are arranged into a segment list in path order by the intelligent scheduling module to form a service path from source to destination; ensuring that service performance requirements are met.

[0111] For step S1013, the intelligent scheduling module extracts the business parameters obtained by the business demand perception module. The business parameters involve real-time control, big data analysis, video rendering, bandwidth requirements, latency sensitivity, periodic task information, and occasional task information.

[0112] Based on the extracted business parameters, combined with the computing node resource information provided by the computing power resource management module and the available bandwidth, current latency, and packet loss rate provided by the network resource management module, a set of candidate computing nodes that meet the basic business requirements is selected. Using the segment routing function based on SRv6 technology, the segment identifiers (SIDs) corresponding to the candidate computing nodes are arranged into a segment list according to the priority order of business requirements, and the priority of each segment is marked in the segment list. The parameters of the segment list are adjusted according to the business cycle characteristics. If the business is a periodic task, resource reservation identifiers for a fixed time period are pre-allocated in the segment list. Here, the parameters of the segment list can be adjusted according to the periodic or sporadic characteristics of the business to ensure that the path can meet both regular requirements and cope with sporadic changes, improving the targeting and reliability of the service path.

[0113] In one embodiment of the present invention, based on step S102, the following is a possible embodiment and its specific implementation will be described in a non-limiting manner. Step S102 specifically includes the following steps:

[0114] Step S1021: The network resource management module scans the nodes in the computing network using the SNMPv3 protocol or the Netconf protocol, and collects the resource information of each node, including the number of CPU cores, memory capacity, number of available cores and supported programming frameworks of computing resources; and the disk type, read and write speed, available space and data redundancy strategy of storage resources.

[0115] Step S1022: Preprocess the collected resource information.

[0116] Preprocessing includes defining a unified resource description template, which converts resources from different manufacturers and models into standardized fields that the system can recognize.

[0117] Step S1023: Assign a globally unique segment identifier (SID) to each processed resource information, and encode the node location and resource type of the resource information into the SID.

[0118] Optionally, node location can be the specific location where the node is set up, or it can be information about the location of the server node, etc. Resource types involve computing, storage, and algorithms.

[0119] Step S1024: Configure the resource information and the corresponding SID to the computing power resource management module, and verify the pre-send status of the resource information through the preset destination address of SRv6: send ICMP probe messages to the node where the resource information is located to detect round-trip latency, packet loss rate and bandwidth utilization; after verification, mark the resource information as available and add it to the computing power resource pool to complete the resource access process.

[0120] As can be seen, this embodiment assigns a SID containing location and type to each resource and establishes a mapping relationship between resources and routing paths; by verifying the actual status of resources, it ensures that resources can be accurately called by the intelligent scheduling module, thereby improving the flexibility and scalability of the resource pool.

[0121] In one embodiment of the present invention, based on step S103, the following will provide a possible embodiment and its specific implementation will be described in a non-limiting manner. For example... Figure 3 As shown, step S103 specifically includes the following steps:

[0122] Step S1031: The business demand perception module collects key characteristic parameters of business traffic in real time, including business type, latency sensitivity, bandwidth requirements, and business cycle, and encapsulates the parameters into demand tags.

[0123] Optionally, the business types include data analysis, data storage, control programs, video processing, etc. Latency sensitivity can range from μs to milliseconds to seconds. Business cycles can include periodic tasks and occasional tasks.

[0124] Step S1032: The intelligent scheduling module calls the interfaces of the computing power resource management module and the network resource management module to obtain the real-time status information of the computing power resource pool, and at the same time extracts the service paths corresponding to each node. The service path is a defined list of segments.

[0125] Real-time status information includes the number of available CPU cores on compute nodes, the amount of free memory, the available IOPS on storage nodes, the remaining bandwidth of network links, current latency, and packet loss rate.

[0126] Step S1033: Match business requirement tags with resource status information: filter resource information that meets latency sensitivity and bandwidth requirements; then, based on node location, generate a scheduling method that includes the target computing node, transmission path, and bandwidth allocation. The scheduling method can be to execute scheduling instructions from the monitoring center node to the corresponding edge computing node and allocate the corresponding bandwidth.

[0127] Step S1034: Verify the feasibility of the scheduling method. The feasibility verification method is to send test messages with service tags to key nodes of the target path to detect the actual latency, packet loss rate and bandwidth availability of the path. If the verification is successful, the policy is sent to the network device to execute traffic scheduling. If the verification fails, the adjustment method is triggered.

[0128] Adjustments can be made by switching to an alternative path or reducing bandwidth allocation for non-critical services, until a feasible strategy is generated.

[0129] It can be seen that by using SRv6 programming instructions to match requirements with resources, key parameters are prioritized and secondary parameters are optimized; specific policies including path and bandwidth allocation are generated, and the feasibility of the policies is ensured by SRv6's path verification function; the verified policies are then sent to network devices for execution, enabling service traffic to be scheduled to the target computing resource pool along the optimal path, thereby improving the adaptability and accuracy of the scheduling policy.

[0130] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0131] The following are embodiments of the intelligent scheduling system for computing power networks provided in this disclosure. This system and the intelligent scheduling methods for computing power networks described above belong to the same inventive concept. Details not fully described in the embodiments of the intelligent scheduling system for computing power networks can be found in the embodiments of the intelligent scheduling methods for computing power networks described above. The system includes:

[0132] The strategy definition module is used to define the access policy and service path of the computing power network in the intelligent scheduling system of computing power network, wherein the intelligent scheduling system of computing power network includes a computing power resource management module, a network resource management module, a business demand perception module, an intelligent scheduling module and a security protection module;

[0133] Computing resources, storage resources, and algorithm resources are connected to the intelligent scheduling system of the computing power network via the network resource management module to form a computing power resource pool;

[0134] Based on the business traffic and application requirements monitored in real time by the business demand perception module, a business scheduling strategy is generated through the programming capabilities of SRv6, and the business traffic is scheduled to the corresponding computing resource pool along the service path.

[0135] According to the service scheduling strategy, the network resource management module creates independent computing network slices in the computing network through network slicing technology, and maintains data transmission within each slice;

[0136] The security protection module establishes an isolation mechanism between the computing power network slices;

[0137] The system monitors the performance of data streams passing through the computing network slices in real time. When an anomaly is detected, it initiates an intelligent operation and maintenance process to troubleshoot and repair the computing network and outputs performance optimization suggestions.

[0138] It should be noted that the computing resource management module can collect and manage various computing resources, storage resources, and algorithm resources in the computing network. By processing resource information and providing it to customers in the form of products, it enables flexible allocation and efficient utilization of resources.

[0139] The network resource management module manages and schedules network resources, including WAN connections, cloud computing nodes, and edge computing nodes, ensuring their rational allocation and efficient utilization. The business demand awareness module captures changes in business traffic and the needs of various applications. It adjusts computing resources to efficiently process and integrate various tasks. The intelligent scheduling module generates business scheduling strategies on demand based on the results of business demand awareness, combined with the status information of computing and network resources. The security protection module constructs an integrated cloud-network-security architecture, achieving comprehensive protection across endpoint security, network security, cloud access security, and cloud security.

[0140] like Figure 4 As shown, this application also provides an electronic device, including a display module 103, a memory 102, a processor 101, and a computer program stored in the memory and executable on the processor 101. When the processor 101 executes the program, it implements the steps of the intelligent scheduling method for computing power networks.

[0141] In embodiments of the present invention, electronic devices include, but are not limited to, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic devices may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments described and / or claimed herein.

[0142] In this embodiment, processor 101 may be implemented using at least one of an application-specific integrated circuit, a programmable logic device, a field-programmable gate array, a processor, a controller, a microcontroller, a microprocessor, or an electronic unit designed to perform the functions described herein. In some cases, such an implementation may be implemented within a controller. For software implementation, implementations such as processes or functions may be implemented with separate software modules that allow the performance of at least one function or operation. Software code may be implemented by a software application (or program) written in any suitable programming language, and the software code may be stored in memory and executed by the controller.

[0143] The display module 103 is used to display information input by the user or information provided to the user. The display module 103 may include a display panel, which may be configured in the form of a liquid crystal display, an organic light-emitting diode, or the like.

[0144] The memory 102 can be used to store software programs and various data. The memory 102 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0145] This application also provides a storage medium storing a computer program thereon, characterized in that the computer program, when executed by a processor, implements the steps of the intelligent scheduling method for the computing power network.

[0146] The storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example,, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0147] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for intelligent scheduling of computing power networks, characterized in that, The methods include: S101: In the intelligent scheduling system for computing power networks, the access policy and service path of the computing power network are defined, wherein the intelligent scheduling system for computing power networks includes a computing power resource management module, a network resource management module, a business demand perception module, an intelligent scheduling module, and a security protection module; Step S101 specifically includes the following steps: S1011: Deploy a computing power network intelligent scheduling system module at a preset node, wherein each module achieves direct interconnection between the physical layer and the data link layer through a high-speed communication interface; S1012: Based on SRv6 technology, each node is assigned a unique segment identifier (SID). The segment identifier contains the node's location information, resource type information, and service level information. Step S1013: Obtain the service type, bandwidth requirements, and latency sensitivity. Utilize the segment routing function of SRv6 to arrange the allocated segment identifiers into a segment list according to the service requirement path order, forming a service path from the service source node to the destination computing node. The segment list configuration method is based on the arrangement of the segment list and the network scheduling's data resource requirements to configure the network scheduling weight. The configuration method of the network scheduling weight is: Zy=α×MG+β×XQ, where MG is the service latency sensitivity, and α and β are the network scheduling's data resource requirements adjustment coefficients. S1014: Verify the validity of the defined service path by sending probe messages to key nodes in the service path to detect the connectivity, latency and packet loss rate of the path, and ensure that the service path meets the minimum performance indicators required by the business. S102: Connect computing resources, storage resources and algorithm resources to the intelligent scheduling system of computing network via the network resource management module to form a computing resource pool; S103: Based on the business traffic and application requirements monitored in real time by the business demand perception module, the intelligent scheduling module generates a business scheduling strategy through the programming capability of SRv6, and schedules the business traffic to the corresponding computing resource pool along the service path. Step S103 specifically includes the following steps: The business demand perception module collects key characteristic parameters of business traffic in real time and encapsulates the parameters into demand tags. The intelligent scheduling module calls the interfaces of the computing power resource management module and the network resource management module to obtain the real-time status information of the computing power resource pool, and at the same time extracts the service path corresponding to each node; Match business requirement tags with resource status information; the matching methods include: filtering resource information that meets latency sensitivity and bandwidth requirements; Based on the node location, a scheduling method is generated that includes the target computing node, transmission path, and bandwidth allocation; The feasibility of the scheduling method is verified by sending test messages with service tags to key nodes of the target path to detect the actual latency, packet loss rate and bandwidth availability of the path. S104: The network resource management module creates independent computing network slices in the computing network through network slicing technology according to the service scheduling strategy, and maintains data transmission within each slice; The network resource management module parses out the service type, resource pool number and service quality level based on the service segment identifier in the service scheduling strategy, and generates network slice information tags accordingly. Based on the network slice information tags, the network resource management module configures a virtual bearer channel on the physical network that uniquely corresponds to the service type; Within the virtual bearer channel, the network resource management module exclusively reserves link bandwidth, cache depth, and forwarding time slots according to the quality of service level, so that different virtual bearer channels on the same physical link form a zero-preemption relationship; When the service scheduling strategy changes, the network resource management module reconfigures the boundaries and reserved resources of the virtual bearer channel in place by configuring incremental fields in the network slice information tag. S105: The security protection module establishes an isolation method between the computing power network slices; The security protection module assigns a slice-level session key derived from the virtual root key to each computing power network slice; The session key K is generated by the virtual root key Kr and the network slice information label D through a one-way function F, where K=F(Kr,D); Insert an isolation tag based on the session key at the entry node of each slice; When the slice lifecycle ends, the security protection module revokes the corresponding session key and broadcasts a key revocation message; During the slice operation, the security protection module periodically rotates the session key, and the new key after rotation remains bound to the same network slice information tag; S106: Monitor the performance of data streams passing through the computing power network slices in real time. When an anomaly is detected, initiate the intelligent operation and maintenance process to troubleshoot and repair the computing power network, and output performance optimization suggestions. Pluggable detection probes are set up along the data flow path of the computing power network slice; After identifying the diagnostic marker, the node records the passing time in the locally maintained diagnostic timestamp table and associates the diagnostic marker with the node's own hardware counter value to form the raw data of hop-by-hop latency and packet loss indication; After receiving a message with raw data, the cloud computing center locates the link segment where the anomaly occurred by comparing the diagnostic timestamps and counter values ​​of adjacent nodes, and generates a compressed repair command containing the anomaly segment identifier and suggested actions. The compressed repair command directly reaches the anomaly segment through the reverse SRv6 path, achieving rapid repair with zero additional routing overhead. After the repair is completed, the cloud computing center records the mapping results of anomaly characteristics and repair strategies in the local performance optimization suggestion library.

2. The intelligent scheduling method for computing power networks according to claim 1, characterized in that, The method includes: Step S102 specifically includes the following steps: The network resource management module scans the nodes in the computing network using the SNMPv3 protocol or the Netconf protocol to collect resource information from each node. The collected resource information is preprocessed; a unique segment identifier (SID) is assigned to each processed resource information, and the node location and resource type of the resource information are configured and encoded into the SID; Configure the resource information and corresponding SID into the computing resource management module, and verify the pre-send status of the resource information through the preset destination address of SRv6: send ICMP probe messages to the node where the resource information is located to detect round-trip latency, packet loss rate and bandwidth utilization; after verification, mark the resource information as available and add it to the computing resource pool to complete the resource access process.

3. A computing power network intelligent scheduling system, characterized in that, The system is used to implement the intelligent scheduling method for computing power networks as described in any one of claims 1 to 2; The system includes: The strategy definition module is used to define the access policy and service path of the computing power network in the intelligent scheduling system of computing power network, wherein the intelligent scheduling system of computing power network includes a computing power resource management module, a network resource management module, a business demand perception module, an intelligent scheduling module and a security protection module; Computing resources, storage resources, and algorithm resources are connected to the intelligent scheduling system of the computing power network via the network resource management module to form a computing power resource pool; Based on the business traffic and application requirements monitored in real time by the business demand perception module, a business scheduling strategy is generated through the programming capabilities of SRv6, and the business traffic is scheduled to the corresponding computing resource pool along the service path. According to the service scheduling strategy, the network resource management module creates independent computing network slices in the computing network through network slicing technology, and maintains data transmission within each slice; The security protection module establishes an isolation mechanism between the computing power network slices; The system monitors the performance of data streams passing through the computing network slices in real time. When an anomaly is detected, it initiates an intelligent operation and maintenance process to troubleshoot and repair the computing network and outputs performance optimization suggestions.

4. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the intelligent scheduling method for computing power networks as described in any one of claims 1 to 2.

5. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent scheduling method for computing power networks as described in any one of claims 1 to 2.

Images