A Multichannel Backup Multicloud Interconnection Method, Device, and Medium Based on a Deterministic Network

Abstract

The present invention provides a multi-path standby multi-cloud interconnection method, device and medium based on a deterministic network, belonging to the technical field of cloud computing. The method includes: constructing a multi-cloud interconnection network architecture to provide deterministic network capabilities for the interconnection of multi-cloud data centers, setting multiple network connection nodes in each cloud data center, and setting multiple standby paths for each connection node; real-time monitoring of the bandwidth, latency, packet loss rate, and jitter parameters of each link; deterministic network path selection and scheduling; selecting the link with the highest network link quality evaluation score for data transmission; through traffic scheduling and dynamic routing mechanisms, adjusting the link selection strategy, intelligently allocating and load balancing the communication traffic of the link, and realizing multi-cloud interconnection. The present invention improves the reliability and business continuity of the network, and provides a more stable and reliable multi-cloud interconnection environment.

Description

Technical Field

[0001] The present invention relates to a method, device, and medium for multi-channel standby multi-cloud interconnection based on a deterministic network, belonging to the technical field of cloud computing. Background Art

[0002] With the rapid development of cloud computing, enterprises and service providers are increasingly inclined to adopt a multi-cloud environment to meet the requirements of high availability, disaster recovery backup, resource elasticity, and performance. Globally, considering security, cost, and efficiency, choosing a multi-cloud approach has become the main form of customers' cloud adoption. The multi-cloud environment allows enterprises to flexibly select different cloud service providers and services according to business needs, thereby optimizing resource allocation and reducing costs. However, in a multi-cloud scenario, what users expect is a technical solution where multiple clouds are physically isolated but can provide a large-bandwidth, low-latency, and low-jitter interconnection network at the network level. The current multi-cloud interconnection technical solutions are mainly implemented based on traditional IP networks, which provide services in a best-effort manner and cannot provide "on-time and accurate" data transmission services. For long-distance and cross-regional multi-cloud integration, traditional IP networks have performance defects such as single-point failures and network congestion, affecting the stability and reliability of the multi-cloud interconnection network.

[0003] With the prosperity of technologies such as cloud computing and 5G / B5G, the emergence of emerging applications such as virtual / augmented reality, smart grids, and remote industrial control requires the network to provide deterministic quality of service. These emerging applications are accompanied by the generation of massive amounts of data, and data processing requires a powerful multi-cloud computing power with cloud-edge-end collaboration and a wide-coverage and high-quality network connection. Computing power applications not only require multi-cloud computing power and a deterministic network, but also need to interconnect multiple clouds scattered in different geographical locations through a deterministic network, and combine and schedule computing tasks to appropriate computing power nodes.

[0004] However, existing multi-cloud interconnection technologies mainly achieve data exchange and resource sharing through network connections between public clouds, private clouds, and hybrid clouds, and have problems such as high data transmission latency, poor reliability, and insufficient scalability. Especially in the face of network failures, there is a lack of an effective backup mechanism to ensure the continuity and security of data. At the same time, considering that enterprises have different requirements for network performance indicators under different business conditions, the requirements for network latency and network jitter indicators for related services such as real-time control and remote monitoring are very high, but the requirements for some office OA, backup systems, etc. are relatively low. The commonly used multi-cloud interconnection methods do not consider the needs for business differences, making it impossible for users to select on-demand among multiple interconnection links with different SLA levels.

[0005] Existing multi-cloud interconnection technology solutions mainly include dedicated leased line connection methods and virtual channel connection methods. The dedicated leased line connection method realizes the interconnection and intercommunication between two clouds by building a dedicated line from one cloud to another. This solution exclusively occupies a physical port on the cloud server computer room equipment and is suitable for scenarios where two clouds are located in the same region and are relatively close. However, it is relatively difficult to achieve interconnection between two clouds across regions and at a relatively long distance, with high costs, and lacks flexibility and scalability. The virtual channel connection method divides relatively independent channels through virtual channel technology on the physical network of operators or other network providers to achieve the interconnection and intercommunication between multiple clouds. In the scenario of realizing cross-regional and long-distance cloud interconnection, this solution requires coordinating the resources of multiple operators, and it is very difficult to control the service quality, making it difficult to meet the requirements of services with high network quality requirements. To sum up, there are many deficiencies in existing multi-cloud interconnection technology solutions and they cannot meet the enterprise's needs for network interconnection with high availability, low latency, high reliability, and bandwidth guarantee. Therefore, a new multi-cloud interconnection method is needed to solve these problems. Summary of the Invention

[0006] The purpose of the present invention is to provide a multi-path backup multi-cloud interconnection method, device, and medium based on a deterministic network, which optimize the network quality between multi-cloud interconnections and improve the transmission efficiency and reliability.

[0007] To achieve the above object, the present invention is realized through the following technical solutions:

[0008] A multi-path backup multi-cloud interconnection method based on a deterministic network includes the following steps:

[0009] Construct a multi-cloud interconnection network architecture to provide deterministic network capabilities for the interconnection and intercommunication of multi-cloud data centers. Set multiple network connection nodes in each cloud data center and set multiple backup paths for each connection node;

[0010] Real-time monitor the bandwidth, latency, packet loss rate, and jitter parameters of each link;

[0011] Deterministic network path selection and scheduling. According to the network topology structure of the service, select a network link quality evaluation method. Select ordinary weighted evaluation for simple topologies, network topology evaluation for medium-complexity topologies, and graph convolutional network evaluation for high-complexity topologies;

[0012] Select the link with the highest network link quality evaluation score for data transmission. If the scores of multiple links are equal, further compare the bandwidth and latency, and select the link with a large bandwidth and low latency;

[0013] Through traffic scheduling and dynamic routing mechanisms, adjust the link selection strategy, intelligently allocate and load balance the communication traffic of the links, and achieve multi-cloud interconnection.

[0014] Preferably, the simple topologies include: point-to-point topology, bus topology; the medium complexity topologies include: star topology, ring topology, tree topology; the high complexity topologies include: mesh topology, dual-ring topology, hybrid topology.

[0015] Preferably, the evaluation formula of the graph convolutional network is as follows:

[0016] ,

[0017] where represents the link quality evaluation score between node and node , represents the activation function, represents the weight of neighbor node , represents the network feature of neighbor node , represents node 's set of neighbor nodes.

[0018] Preferably, the traffic scheduling and dynamic routing mechanism adopts the shortest path calculation and broadband-constrained path method. By calculating the shortest path from the source node to other nodes, it judges whether the shortest path meets the broadband constraint, and selects the shortest path that meets the broadband constraint for load balancing.

[0019] Preferably, the shortest path calculation formula is as follows:

[0020] ,

[0021] where represents the shortest path from node to node , represents the bandwidth or delay cost from node to node , represents the set of edges in the topological structure.

[0022] Preferably, the broadband constraint is as follows: the minimum cost between node and node should satisfy that the total traffic between node and node must be less than or equal to the physical link bandwidth between node and node .

[0023] A multi-path standby multi-cloud interconnection device based on a deterministic network, comprising a processor and a memory storing program instructions, the processor being configured to execute the multi-path standby multi-cloud interconnection method based on a deterministic network when running the program instructions.

[0024] A computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the multi-path standby multi-cloud interconnection method based on a deterministic network.

[0025] The advantages of the present invention are as follows: By introducing the idea of a deterministic network and combining a multi-dimensional link evaluation and weight calculation model, the present invention realizes the selection of multiple standby paths and an automatic fault switching mechanism. This mechanism ensures that when the main network path fails, it can quickly and automatically switch to the standby path, thus significantly improving the reliability and business continuity of the network. Compared with traditional static routing switching or manual intervention-based fault recovery methods, the automatic fault detection and switching mechanism of the present invention greatly shortens the recovery time and reduces the possibility of service interruption, providing a more stable and reliable multi-cloud interconnection environment for enterprises and service providers.

[0026] Traditional cross-cloud interconnection solutions often rely on the Internet or conventional virtual private networks (VPNs), and these network connections are uncertain, easily leading to problems such as bandwidth fluctuations and unstable delays. However, through a deterministic network, the present invention can precisely control network performance parameters such as bandwidth, delay, and packet loss rate, providing stable and predictable network performance for cross-cloud communication. Especially in terms of delay and bandwidth guarantee, the present invention is significantly superior to traditional solutions, meeting the high-performance service requirements such as real-time control and remote monitoring.

[0027] The present invention introduces a traffic scheduling and dynamic routing mechanism, which can select the optimal path according to real-time network conditions, link loads, bandwidth requirements, etc., avoiding network path congestion and imbalance. This intelligent scheduling based on network status greatly improves the throughput and efficiency of multi-cloud interconnection, ensuring the smooth operation of services. At the same time, it also provides a more flexible and efficient way of using network resources for enterprises and service providers. Description of the Drawings

[0028] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention and do not constitute a limitation to the present invention.

[0029] Figure 1 It is a schematic diagram of the method flow of the present invention. Detailed Embodiments

[0030] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0031] Embodiment 1

[0032] As Figure 1 shown, a multi-path standby multi-cloud interconnection method based on a deterministic network includes the following steps:

[0033] S1: Construct a multi-cloud interconnection network architecture to provide deterministic network capabilities for the interconnection of multi-cloud data centers. Set multiple network connection nodes in each cloud data center, and set multiple standby paths for each connection node.

[0034] As a refinement of the above embodiment, this embodiment constructs a multi-cloud interconnection architecture that supports multi-path standby. This solution can provide interconnection and intercommunication between all cloud data centers with deterministic network capabilities. A deterministic network is a network architecture that can provide predictable and strictly guaranteed network performance. Compared with traditional network architectures (such as IP-based networks), a deterministic network optimizes the network through technologies such as high-precision clock synchronization, multiple transmission and selection reception, bandwidth reservation and resource guarantee, and intelligent routing at the IP layer. It not only pays attention to the transmission reliability of data, but also particularly emphasizes the strict guarantee of performance indicators such as delay, bandwidth, and packet loss rate during the data transmission process. A deterministic network ensures that network traffic is transmitted in a predetermined manner by adopting advanced traffic scheduling and control methods, combined with technologies such as network slicing, time-sensitive networking (TSN), and software-defined networking (SDN), and can accurately predict and regulate network behavior. Moreover, a deterministic network has an advantage that it can dynamically adjust the proportion of deterministic bandwidth slices according to the actual load requirements of the service. Multiple network connection nodes are set in each cloud data center, and each node can connect to different cloud platforms or communicate internally through a dedicated virtual network. To improve reliability, multiple standby paths are set for each connection node.

[0035] S2: Real-time monitor the bandwidth, delay, packet loss rate, and jitter parameters of each link.

[0036] As a refinement of the above embodiment, the controller periodically sends probe packets to each link to measure the real-time state of the link.

[0037] S3: Deterministic network path selection and scheduling. Select the network link quality evaluation method according to the network topology structure of the service. For simple topologies, select the ordinary weighted evaluation; for medium-complexity topologies, select the network topology evaluation; for high-complexity topologies, select the graph convolutional network evaluation.

[0038] As a refinement of the above embodiment, in order to fully evaluate the quality of network links, this embodiment introduces a multi-dimensional and multi-method link evaluation and weight calculation method. By adopting three different-dimensional weight evaluation methods of ordinary weighting, network topology, and graph convolutional network, each evaluation method has its specific application scenario, and the network performance indicators can be calculated more scientifically and accurately according to business requirements, providing better deterministic capabilities for multi-cloud interconnection links.

[0039] The simple topologies include: point-to-point topology, bus topology; the medium-complexity topologies include: star topology, ring topology, tree topology; the high-complexity topologies include: mesh topology, dual-ring topology, hybrid topology.

[0040] The ordinary weighted evaluation method is suitable for monitoring and evaluating common metrics such as bandwidth and delay in simple network topologies, and has poor adaptability to complex network topologies. It may not be able to capture some potential network bottlenecks or complex delay changes. Some applicable scenarios and parameter suggestions in this embodiment are listed as follows (for example only, these parameters can be adjusted and configured by users in the controller interface template parameters):

[0041] Services sensitive to bandwidth and network packet loss: For some applications that require strict control of bandwidth and network packet loss, such as high-definition video streaming, real-time voice communication, etc., the ordinary weighted evaluation method based on network metrics can effectively combine network metrics such as bandwidth and packet loss rate to select the most suitable path.

[0042] Simple application scenarios: This method is suitable for scenarios with relatively simple network structures and not very complex network topologies, because it relies on traditional network parameters (such as bandwidth, delay, etc.) for evaluation, and the weighting strategy can be adjusted according to the specific requirements of the service.

[0043] Conventional services: Such as general enterprise internal applications, standard Web access, etc., in scenarios where the requirements for network quality are relatively balanced. In these scenarios, the weighted evaluation of network metrics can provide a more intuitive network quality evaluation.

[0044] Optionally, the network topology evaluation method mainly uses the network topology structure (such as the connection relationship between nodes and edges) and link delay (i.e., the communication delay between different links) to evaluate the network quality. This method determines the availability and optimality of different paths by considering the link delay. This method can accurately capture the impact of different links and network topologies on the delay, is applicable to multi-region and complex network environments, and can help identify links and paths with high delays. At the same time, the requirements of the current algorithm for computing resources also need to be considered, because accurate network topology and link delay data are required, and a large amount of computing and data collection are needed for large-scale networks. Some applicable scenarios are listed in this embodiment as follows:

[0045] Interconnection of cloud data centers under large-scale networks: In complex multi-cloud environments and large-scale data center networks, the impact of network topology cannot be ignored, and link delay may have an important impact on the overall network performance. Therefore, the evaluation method based on topology and link delay is applicable to large-scale cloud network interconnection scenarios that require optimizing path delay.

[0046] Delay-sensitive real-time applications: For applications that are extremely sensitive to delay, such as real-time video conferencing, online games, financial transactions, etc., optimizing link delay is crucial. In these scenarios, evaluating link delay and network topology structure helps to select low-delay paths, thereby enhancing the business experience.

[0047] Interconnection of wide-area multi-clouds in different locations: When transferring data between different cloud service providers, due to different geographical locations, the changes in network topology and link delay are very significant. Therefore, the evaluation method based on topology and delay can optimize the interconnection efficiency across regions and clouds.

[0048] Optionally, the evaluation formula of the graph convolutional network is as follows:

[0049] ,

[0050] where, represents the link quality evaluation score between node and node . represents the activation function, represents the weight of neighbor node . represents the network feature of neighbor node . represents the set of neighbor nodes of node .

[0051] This method is a neural network applicable to graph-structured data and has great application potential in link quality assessment. By modeling the network topology as a graph structure, it can more effectively capture the dependencies between links, thereby optimizing the assessment of link quality and weight allocation. Through this assessment method, on the basis of considering the network topology structure, the mutual influence and time-variability between links are comprehensively considered. Using the information of graph nodes (network devices) and edges (links), the global structure and local attributes of the network are learned through convolution operations, so as to conduct network quality assessment. Its innovation lies in being able to handle the link quality assessment problem in complex networks by learning the local features and global structure of the graph.

[0052] This method can handle complex non-Euclidean data (such as network topology), adapt to dynamic network environments, has the ability of self-learning and self-adaptation, can optimize the assessment results according to historical data, and is applicable to large-scale and multi-dimensional network performance optimization. In order to accurately assess network quality, this method requires a large amount of training data and computing resources. Some applicable scenarios are listed in this embodiment as follows:

[0053] Complex network topologies and large-scale networks: GCN can handle complex network topologies and help identify potential bottlenecks and optimization paths in the network by learning the relationships between nodes and edges in the network structure. It is applicable to large-scale and dynamically changing network environments, such as large-scale enterprise networks or multi-cloud interconnection architectures.

[0054] Intelligent path selection and optimization: For scenarios that require dynamically selecting the optimal path in a multi-cloud environment, GCN can predict the most suitable path by learning historical data. It is especially applicable to scenarios that require self-adaptation and intelligent optimization, such as intelligent manufacturing, Internet of Things (IoT), etc.

[0055] Automated network operation and maintenance network: In network operation and maintenance, GCN can automatically model the network topology and status, and real-time evaluate and predict network performance, which is particularly applicable to scenarios such as automated fault detection, optimized traffic scheduling, and resource allocation.

[0056] S4: Select the link with the highest network link quality assessment score for data transmission. If the scores of multiple links are equal, further compare the bandwidth and latency, and select the link with a large bandwidth and low latency.

[0057] As a refinement of the above embodiment, the main path selection is based on the comprehensive quality score Qlink of each link. The controller will select the link with the highest score as the main path. If there are multiple links with similar scores, the bandwidth requirements and application types of the links will be further considered, and the link with a larger bandwidth and lower latency will be preferentially selected. To avoid the problem of a single path bottleneck, a backup path is selected according to indicators such as link redundancy and load balancing. When a certain path fails or does not meet the performance requirements, the system will automatically switch to the backup path to ensure the reliability of the network. The system dynamically adjusts the path selection strategy by real-time monitoring the network status, thereby maximizing the stability and performance of data transmission.

[0058] S5: Through the traffic scheduling and dynamic routing mechanism, adjust the link selection strategy, intelligently allocate and balance the communication traffic of the links, and achieve multi-cloud interconnection.

[0059] As a refinement of the above embodiment, the present invention realizes the intelligent allocation and load balancing of communication traffic between different clouds through the traffic scheduling and dynamic routing mechanism. The system adjusts the path selection strategy when the service is issued according to the network load situation monitored in real time. The core of load balancing is to use an algorithm to regularly detect the performance of each link and dynamically select the most suitable path according to the current state of the link to ensure the reasonable utilization of network resources. For example, if the bandwidth utilization rate of a certain link is close to 100%, the system will actively select a backup link for traffic sharing to avoid performance degradation caused by link overload. This mechanism can dynamically adjust the traffic direction according to factors such as network load, link status, and bandwidth requirements, avoid congestion on a single path, and improve the overall network performance. The routing scheduling in the present invention is calculated and evaluated by using the method of shortest path calculation and bandwidth-constrained path. The specific algorithm is as follows:

[0060] The formula for the shortest path is as follows:

[0061] ,

[0062] Among them, represents the shortest path from node to node , represents the bandwidth or latency cost from node to node , represents the set of edges in the topological structure.

[0063] The broadband constraint is as follows: The minimum cost between node and node should satisfy that the total traffic between node and node must be less than or equal to the sum of the traffic between node and node The bandwidth of the physical link between. There may be multiple service traffic flows between two device nodes, so the sum of the bandwidths of multiple service traffic flows cannot exceed the bandwidth value of the physical link.

[0064] Specifically, it can be expressed as:

[0065] ,

[0066] Among them, represents on the premise of a certain condition, represents a certain actual service traffic flow between node and node , represents node and node the maximum bandwidth value of the physical link between, represents the set of nodes in the network link topology, represents node and node the link between exists as a valid link in the current network topology.

[0067] Based on the bandwidth-constrained path method, when the service is issued between nodes u and v, it will be judged in a dry-run manner to ensure that the bandwidth requirements of the new service are within the bandwidth limit, avoiding the situation of link overload between nodes u and v after the service is successfully issued.

[0068] It should be noted that: This application proposes a multi-path backup multi-cloud interconnection method based on Deterministic Networking (DetNet), innovatively applying the idea of deterministic networks to the multi-cloud interconnection scenario. Deterministic networks solve problems such as high latency and bandwidth fluctuations caused by unstable paths in traditional Internet and Virtual Private Networks (VPNs) by providing guarantees for key performance indicators such as predictable bandwidth, latency, and packet loss rate for network traffic. This invention innovatively uses multiple backup paths for the interconnection of multi-cloud data centers. When the primary path fails, the system can quickly switch to the backup path, ensuring the high availability and reliability of the network connection.

[0069] By introducing a traffic scheduling and dynamic routing mechanism, this application can intelligently adjust routing and traffic allocation based on real-time information such as network status, link load, bandwidth requirements, and fault detection. Different from traditional static routing and traffic allocation methods, the intelligent routing and traffic scheduling mechanism of the present invention can make rapid adjustments according to real-time network load and performance requirements, ensuring optimal transmission performance and bandwidth utilization efficiency under different network conditions. This dynamic traffic scheduling and load balancing strategy helps to avoid network bottlenecks, reduce latency, and improve the overall network throughput and stability.

[0070] The multi-path backup path mechanism of this application not only supports redundancy at the physical layer and link layer, but also can achieve redundancy of logical paths based on the virtual network layer between clouds, providing higher reliability for cross-cloud traffic between multiple clouds. Especially in scenarios involving cross-cloud disaster recovery and highly reliable services, the solution of the present invention can effectively avoid single-point failures and improve the robustness of multi-cloud interconnection.

[0071] Embodiment 2

[0072] An embodiment of the present disclosure provides a multi-path backup multi-cloud interconnection device based on a deterministic network, including a processor and a memory. Optionally, the device may further include a communication interface and a bus. Among them, the processor, communication interface, and memory can complete communication with each other through the bus. The communication interface can be used for information transmission. The processor can call the logical instructions in the memory to execute the multi-path backup multi-cloud interconnection method based on the deterministic network in the above embodiment.

[0073] In addition, when the logical instructions in the above memory are implemented in the form of a software functional unit and sold or used as an independent product, they can be stored in a computer-readable storage medium.

[0074] As a computer-readable storage medium, the memory can be used to store software programs and computer-executable programs, such as program instructions / modules corresponding to the methods in the embodiments of the present disclosure. The processor executes functional applications and data processing by running the program instructions / modules stored in the memory, that is, implements the multi-path backup multi-cloud interconnection method based on the deterministic network in the above embodiments.

[0075] The memory may include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the terminal device, etc. In addition, the memory may include high-speed random access memory and may also include non-volatile memory.

[0076] An embodiment of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, which are configured to execute the above-described multi-path standby multi-cloud interconnection method based on a deterministic network.

[0077] The above computer-readable storage medium may be a transient computer-readable storage medium or a non-transient computer-readable storage medium.

[0078] The technical solution of the embodiment of the present disclosure may be embodied in the form of a software product. The computer software product is stored in a storage medium and includes one or more instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the embodiment of the present disclosure. The foregoing storage medium may be a non-transient storage medium, including: various media that can store program codes such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, or may also be a transient storage medium.

[0079] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art may still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multi-path backup multi-cloud interconnection method based on a deterministic network, characterized in that: The following steps are involved: Build a multi-cloud interconnection network architecture to provide deterministic network capabilities for the interconnection of multi-cloud data centers. Set up multiple network connection nodes in each cloud data center and set up multiple backup paths for each connection node. Real-time monitoring of bandwidth, delay, packet loss rate, and jitter parameters of each link; Deterministic network path selection and scheduling. Select the network link quality assessment method based on the network topology of the business. For simple topology, select ordinary weighted assessment. For medium-complexity topology, select network topology assessment. For high-complexity topology, select graph convolutional network assessment. The link with the highest network link quality assessment score is selected for data transmission. If multiple links have equal scores, the bandwidth and delay are further compared to select the link with large bandwidth and low delay. Through traffic scheduling and dynamic routing mechanisms, the link selection strategy is adjusted, the communication traffic of the link is intelligently allocated and load balanced, and multi-cloud interconnection is achieved.

2. The method for interconnecting multiple clouds with multiple backup channels based on a deterministic network according to claim 1, characterized in that: The simple topologies include: point-to-point topology and bus topology; the medium-complexity topologies include: star topology, ring topology and tree topology; the high-complexity topologies include: mesh topology, dual-ring topology and hybrid topology.

3. The multi-path backup multi-cloud interconnection method based on a deterministic network according to claim 1, characterized in that: The graph convolutional network evaluation formula is as follows: ， in, Representation Node and nodes The link quality assessment score between represents the activation function, Represents neighbor nodes The weight of Represents neighbor nodes The network characteristics of Representation Node The set of neighbor nodes.

4. The method for interconnecting multiple clouds with multiple backup channels based on a deterministic network according to claim 1, characterized in that: The traffic scheduling and dynamic routing mechanism adopts the shortest path calculation and broadband constraint path method, calculates the shortest path from the source node to other nodes through the shortest path, determines whether the shortest path meets the broadband constraint, and selects the shortest path that meets the broadband constraint for load balancing.

5. The method for interconnecting multiple clouds with multiple backup channels based on a deterministic network according to claim 4, characterized in that: The shortest path calculation formula is as follows: ， in, Representation Node To Node The shortest path of Represents a slave node To Node bandwidth or latency costs, Represents a set of edges in a topological structure.

6. The multi-path backup multi-cloud interconnection method based on a deterministic network according to claim 5, characterized in that: The bandwidth constraints are as follows: and nodes The minimum cost must satisfy the slave node and nodes The total traffic between nodes must be less than or equal to and nodes The bandwidth of the physical link between them.

7. A multi-path backup multi-cloud interconnection device based on a deterministic network, comprising a processor and a memory storing program instructions, characterized in that: The processor is configured to execute the multi-path backup multi-cloud interconnection method based on a deterministic network as described in any one of claims 1-6 when running the program instructions.

8. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the program is executed by a processor, a multi-path backup multi-cloud interconnection method based on a deterministic network is implemented as described in any one of claims 1 to 6 above.

Images