A routing architecture and method for a computing power service open internet

Abstract

This invention relates to the field of network technology, specifically to a routing architecture and method for an open interconnected network for computing power services. The architecture includes: a network access device; a network output device; a network control center; and a network management device, used to generate a first orchestration result between the first network output device and a target service provider node based on parsing results, computing resource status, and network resource status. The network management device is also used to distribute the first orchestration result to the network access device and send service request packets to the wide area network routing node, so that the service request packets are sent to the first network output device according to the specified path and strategy in the first orchestration result. The first network output device is used to send the service request packets to the target service provider node. By implementing the technical solution of this invention, more refined, flexible, and reliable end-to-end routing control is provided, achieving full-process protection of transmission and computing and optimal utilization of network and computing resources.

Description

Technical Field

[0001] This invention relates to the field of network technology, and specifically to a routing architecture and method for an open interconnection network for computing power services. Background Technology

[0002] As a channel for data transmission and information delivery, the network plays an indispensable and fundamental role in the digital economy. Since the birth of the internet, the evolution of network technology has been inextricably linked to the changing needs of applications. Early computer applications, such as email and file transfer, had relatively simple functions, focusing only on accessibility and connectivity. Therefore, the network merely provided communication connections for applications, maintaining relative isolation between the network and applications. With the emergence of new models and business forms such as streaming media and instant messaging, the demands of applications for mobility, real-time performance, and the Internet of Things spurred the development of mobile internet technologies such as 3G / 4G / 5G, further enhancing the network's communication capabilities. In recent years, the rapid development of new technologies and businesses such as artificial intelligence, big data, and cloud computing has led to an evolution in the communication paradigm between internet applications that support these advancements, from the traditional one-stage "request-retrieve" model to a more complex two-stage "request-computation-retrieve" model. This new communication paradigm requires the network to not only provide general communication channels but also to incorporate the computational process into its control. With the ubiquitous supply of computing power and the containerization and service-oriented development of computing architecture, computing resources are gradually shifting from closed private to open public. This means that the network needs to take on the role of opening up computing to the outside world in the form of common Internet services. It also guides network technology innovation to shift from host interconnection to service interconnection, thereby deepening the integration of computing and network and promoting the development of the digital economy in society.

[0003] However, due to the current privatization of computing power and service resources by service providers, the existing network architecture and protocols are difficult to openly share. Furthermore, the future development of computing networks involves the deep integration of computing power and networks, requiring new network architectures and large-scale infrastructure construction for some theories and technologies, which is difficult to achieve in this way. Secondly, user demands for computing power and services are becoming increasingly homogenized; users may require different service providers for different service needs, necessitating user selection. Thirdly, while service layer routing deployment is simple and efficient, greatly simplifying the routing process, the lack of awareness and control over network resources means this mechanism can only provide coarse-grained performance guarantees. Furthermore, the state of computing power and resources in computing networks is highly real-time. Currently, the state of service provider computing power and resources is difficult to update in real time, and there are significant differences in the categories and stability of computing and network resource states. For example, the state of computing resources, especially service instance resources, is highly dynamic, potentially experiencing millisecond-level state change frequencies in extreme cases. This inconsistency in computing and network resource characteristics inevitably leads to serious derivative problems such as routing table bloat and routing oscillations. Summary of the Invention

[0004] In view of this, the present invention provides a routing architecture and method for an open interconnection network for computing power services, in order to solve the problems of service interconnection, sharing and deployment caused by different network facilities, architectures and data security issues of service providers, as well as the need for large-scale transformation of existing infrastructure by new technologies.

[0005] In a first aspect, the present invention provides a routing architecture for an open interconnection network for computing power services, comprising: a network access device for receiving service request messages from user terminals and for forwarding service request messages to a network management device; a network output device for receiving computing resource status of each service providing node sent by a server and for forwarding computing resource status to the network management device; a network control center for notifying the network management device of the network resource status of each wide area network routing node; wherein the network resource status includes the network topology and resource status between each wide area network routing node; and a network management device for parsing service request messages to obtain parsing results; when the network management device senses the basic... During the operation of the infrastructure layer, the network management device is also used to generate a first orchestration result between the first network output device and the target service provider node based on the parsing result, computing resource status, and network resource status. The first orchestration result includes hop-by-hop routing information corresponding to the service request message, which is generated based on the network resource status. The network management device is also used to distribute the first orchestration result to the network access device. The network access device is also used to send the service request message to the WAN routing node, so that the service request message is sent to the first network output device according to the specified path and strategy in the first orchestration result. The first network output device is used to send the service request message to the target service provider node according to a preset routing strategy.

[0006] The routing architecture of the open interconnection network for computing power services provided in this embodiment of the invention requires the intelligent management module to have comprehensive management capabilities through "service layer + infrastructure layer collaboration" routing. However, it also supports more refined, flexible, and reliable end-to-end routing control, achieving full-process protection of transmission and computing and optimal utilization of network and computing resources. In other words, the system can allocate resources more accurately and improve the utilization rate of network and computing resources. By parsing service request packets through network management devices and generating hop-by-hop routing information based on network resource status, the system can effectively manage the network topology and resource status between WAN routing nodes. This helps optimize the allocation and use of network resources and solve the interconnection problem between different network facilities and architectures. Through the functions of the network management devices, the system can generate and distribute preset routing policies to securely send service request packets to the target service provider node. This secure routing policy helps service providers ensure the security and integrity of data during transmission, thereby solving data security issues. Through the optimized routing scheme and resource management of this system, existing infrastructure can be utilized more effectively, the transformation costs brought by new technologies can be reduced, and the smooth interconnection, sharing, and deployment of services can be ensured.

[0007] In one optional implementation, the routing architecture of the open interconnection network for computing power services further includes: when the network management device cannot perceive the operating status of the infrastructure layer, the network management device generates a second orchestration result between the network access device and the second network output device based on the parsing result and the computing resource status; the network management device is also used to send the second orchestration result to the network access device; the network access device is also used to send service request messages to the wide area network routing node so that the service request messages are sent to the second network output device according to the specified path and strategy in the second orchestration result.

[0008] The routing architecture of the open interconnection network for computing power services provided in this invention enables smooth network evolution based on existing network infrastructure. This means the system can be upgraded gradually without impacting existing network facilities, ensuring service continuity and stability. Users only need to provide specific requirements, eliminating the need to select service providers. This simplifies the user access process, improves user experience, and reduces operational difficulty. The two-layer routing mechanism—service layer routing and "service layer + infrastructure layer collaborative" routing—each has its advantages and disadvantages but complements each other, providing a feasible approach for large-scale network deployment and optimizing large-scale network deployment through complementary methods.

[0009] Secondly, the present invention provides a routing method for an open interconnection network for computing power services, applied to the routing architecture of the open interconnection network for computing power services described in the first aspect above, comprising: a user terminal initiating a service request message and reporting the service request message to a network management device through a network access device; a server sending the computing resource status of each service providing node to the network management device through a network output device; a network control center notifying the network management device of the network resource status of each wide area network routing node; the network management device parsing the service request message to obtain the computing power requirement and network bandwidth requirement corresponding to the service request message; when the network management device cannot perceive the operating status of the infrastructure layer, the network management device generates a second orchestration result between the network access device and the second network output device based on the computing power requirement and computing resource status; the network management device then... The second orchestration result is sent to the network access device; the network access device sends the service request message to the WAN routing node so that the service request message can be sent to the second network output device according to the specified path and strategy in the second orchestration result; when the network management device senses the operating status of the infrastructure layer, the network management device generates a first orchestration result between the first network output device and the target service provider node based on computing power requirements, network bandwidth requirements, computing resource status, and network resource status; the network management device sends the first orchestration result to the network access device; the network access device sends the service request message to the WAN routing node so that the service request message can be sent to the first network output device according to the specified path and strategy in the first orchestration result; the first network output device sends the service request message to the target service provider node according to the preset routing strategy.

[0010] The routing method for open interconnected computing power services provided in this invention simplifies the user process by allowing users to provide specific requirements without needing to filter service providers. This reduces management and selection costs, enabling users to focus more on their needs and improving user experience. This method enables smooth network evolution based on existing network infrastructure, effectively integrating with existing infrastructure and minimizing major modifications, thus reducing deployment and upgrade risks and costs. The two-layer routing mechanism—service layer routing and "service layer + infrastructure layer collaborative" routing—complements each other, supporting more granular, flexible, and reliable end-to-end routing control. It ensures routing stability and the timeliness of measurement information, maintaining routing stability while promptly acquiring and utilizing the latest network and computing resource statuses. The intelligent management module performs primary routing between network access devices and network output devices based on computing power requirements and computing resource status, and secondary routing between network output devices and target service provider nodes based on computing power requirements, network bandwidth requirements, computing resource status, and network resource status. This achieves full-process protection of transmission and computing and optimal utilization of network and computing resources. This comprehensive resource scheduling and routing control helps improve resource utilization and overall performance.

[0011] In an optional implementation, when the network management device senses the operating status of the infrastructure layer, the method further includes: the user terminal generating a corresponding identifier locally based on the type of the service request content, and encapsulating the identifier into a service request message to obtain a first service request message; after receiving the first service request message, the network access device using the identifier as an index to query locally deployed routing entries, and encapsulating the corresponding routing information into the first service request message to obtain a second service request message; reporting the second service request message to the network management device; the network management device parsing the second service request message to obtain the computing power requirement and network bandwidth requirement corresponding to the service request message; and generating a third service request message based on the computing power requirement, network bandwidth requirement, computing resource status, and network resource status. A first orchestration result is established between a network output device and a target service provider node; the first orchestration result is then distributed to a network access device; the network access device, based on the first orchestration result, sends a second service request message to a WAN routing node for routing forwarding; the WAN routing node sends the second service request message to the first network output device according to the path and policy specified in the first orchestration result; the first network output device parses the second service request message to obtain an identifier; based on the identifier, it selects a target service provider node, replaces the original Internet Protocol address and the corresponding original port number of the application in the second service request message with the target Internet Protocol address and target port number of the target service provider node, and transmits the replaced second service request message to the target service provider node.

[0012] The routing method for open interconnected computing power services provided in this invention involves user terminals generating identifiers locally and encapsulating them in service request messages. Network access devices can directly use these identifiers to query locally deployed routing entries, thereby reducing reliance on network management devices and improving the efficiency and response speed of route resolution. After sensing the operational status of the infrastructure layer, the network management device generates dynamic orchestration results based on the type of service request content, computing power requirements, network bandwidth requirements, and real-time computing and network resource status. This enables intelligent routing and dynamic adjustment of service requests, improving resource utilization and network performance. The identifier generated by the user terminal remains unchanged throughout the service request process. Selecting the target service provider node through the identifier achieves transparent processing of service requests, improving network manageability and maintainability. The routing information encapsulated in the first service request message can be customized according to the type of service request content, enabling the network to better adapt to different types of service requests and provide more personalized services. After the first network output device parses the service request message and replaces the original Internet Protocol address and port number, secure processing of the service request can be achieved, ensuring the legitimacy and security of the service request and preventing malicious attacks and unauthorized access.

[0013] In one optional implementation, the network management device is used to generate a second orchestration result between the network access device and the second network output device based on computing power requirements and computing resource status, including: the network management device generates a second orchestration result between the network access device and the second network output device based on a preset service layer routing model, computing power requirements, and computing resource status.

[0014] The routing method for open interconnected computing power services provided in this invention optimizes resource utilization and avoids waste by orchestrating network devices according to computing power requirements and resource status. A preset service layer routing model ensures network service quality and stability, improving user experience. Network management devices automatically adjust the orchestration of network devices based on the preset model and real-time status, reducing manual intervention and management costs. Automated network device orchestration improves network operation efficiency and response speed, reducing the likelihood of failures.

[0015] In one optional implementation, the network management device generates a first orchestration result between the first network output device and the target service provider node based on computing power requirements, network bandwidth requirements, computing resource status, and network resource status. This includes: the network management device generating the first orchestration result between the first network output device and the target service provider node based on a preset service layer and infrastructure layer collaborative routing model, computing power requirements, bandwidth requirements, computing resource status, and network resource status.

[0016] The routing method for open interconnected computing power services provided in this invention, by considering factors such as computing power requirements, network bandwidth requirements, computing resource status, and network resource status, enables network management devices to make intelligent decisions, thereby more effectively allocating resources and optimizing network performance. Through a pre-defined service layer and infrastructure layer collaborative routing model, network devices and resources at different levels can be better coordinated, improving the overall efficiency and reliability of the network. By considering factors such as computing power and bandwidth requirements, resource utilization can be better optimized, avoiding resource waste and bottlenecks, thereby improving network performance and scalability. By rapidly generating orchestration results, network management devices can make timely adjustments and optimizations to adapt to dynamic network changes and ever-increasing demands, thus maintaining network stability and reliability.

[0017] In one optional implementation, the computing power requirement includes service type requirement, computing power capability requirement, and latency guarantee requirement; the service type requirement is the service type of the specific computing power service corresponding to the service request message; the computing power capability requirement is the computing power requirement size of the specific computing power service; and the latency guarantee requirement is the completion time limit corresponding to the specific computing power service.

[0018] In one alternative implementation, the service layer routing model is represented as follows:

[0019] max(min){f(S C )}

[0020]

[0021]

[0022]

[0023]

[0024] in, S C It indicates the real-time computing resource status of each service provider node; SID represents the various services in the routing architecture of the open interconnection network for accessing computing power services; Indicates the egress gateway e mAll service-providing nodes within the local area network of the server provide SIDs. n The sum of computing resources used by the corresponding services; f(S) C This represents the evaluation metric for the primary routing orchestration results, using... These represent the ingress gateway set, egress gateway set, and SID set, respectively.

[0025] In one alternative implementation, the service layer and infrastructure layer collaborative routing model is represented as follows:

[0026] max(min){g(S C ,S B )}

[0027]

[0028]

[0029]

[0030]

[0031]

[0032] in, S B This indicates the real-time bandwidth resource status of each WAN routing node; Represents a WAN routing node r m to r n The bandwidth occupied by a unidirectional link; if there is no direct link between the two nodes, the element value is 0; g(S C ,S B ) represents the evaluation index for the secondary routing arrangement results.

[0033] In an optional implementation, if the WAN routing nodes and direct links have time deterministic guarantee capabilities, the service layer and infrastructure layer collaborative routing model can be extended to a service layer and infrastructure layer collaborative deterministic routing model to achieve determinism in the time dimension.

[0034] The collaborative deterministic routing model between the service layer and the infrastructure layer is represented as follows:

[0035] max(min){g(S C ,S B )}

[0036]

[0037]

[0038]

[0039]

[0040] The routing method for open interconnected computing power services provided in this invention, through a cooperative deterministic routing model, enables the network to provide time deterministic guarantees, ensuring that the transmission time of data packets within the network is controllable. For services requiring time determinism, the cooperative deterministic routing model can effectively improve service quality, ensuring that data packets arrive at their destination on time, reducing data transmission latency and volatility, and enhancing user experience. By considering time-dimensional determinism in the routing model, the network system can more accurately plan and manage data transmission paths, reducing the likelihood of network congestion and failures, thereby enhancing system reliability and stability. Attached Figure Description

[0041] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0042] Figure 1 This is a schematic diagram of the routing architecture of the open interconnection network for computing power services according to an embodiment of the present invention;

[0043] Figure 2 This is a schematic diagram of a primary routing method according to an embodiment of the present invention;

[0044] Figure 3 This is a schematic diagram of a secondary routing system according to an embodiment of the present invention;

[0045] Figure 4 This is a flowchart illustrating the routing method of the open interconnection network for computing power services according to an embodiment of the present invention;

[0046] Figure 5 This is a schematic diagram of two-layer routing according to an embodiment of the present invention;

[0047] Figure 6 This is a flowchart illustrating another routing method for an open interconnected computing service network according to an embodiment of the present invention;

[0048] Figure 7 This is a flowchart illustrating another routing method for an open interconnected computing power service network according to an embodiment of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0050] In a service-centric network, computing power is the most crucial resource. Currently, computing power resources are primarily concentrated in the private networks of various service providers. Given the existing network infrastructure, architecture, and protocols of these providers, and considering factors such as data security, interconnecting and sharing computing power resources is difficult. Therefore, a new type of network is needed to support this, enabling services to be shared and computing power resources to be made available.

[0051] When discussing new types of networks, "stack clearing" and "evolution" can be considered as two different paths or methods for thinking about the development of network architecture or network technology. While stack clearing can more thoroughly realize technological innovation and network transformation, years of research have shown that it is difficult to achieve. Current network infrastructure is massive in scale, and the existing network architecture, protocols, and other technologies and solutions are nearing perfection. On the other hand, service providers such as data center services, cloud services, and enterprise network services have reached a significant scale. Evolution, on the other hand, only gradually improves and refines the network based on the existing network architecture. This includes introducing new technologies, upgrading existing protocols, improving network security, and enhancing performance, while preserving existing infrastructure and technologies as much as possible. Evolution is typically more gradual, allowing the network to continuously adapt to changing needs without requiring fundamental reconstruction.

[0052] With the rapid development of new technologies and services such as artificial intelligence, big data, and cloud computing, the communication paradigm between internet applications, which plays a supporting role in these advancements, has evolved from the traditional one-stage "request-retrieve" model to a more complex two-stage "request-computation-retrieve" model. This new communication paradigm not only requires the network to provide traditional channels for data transmission and information delivery, but also implies that the network needs to assume the role of opening up computing as a public internet service, thereby deepening the integration of computing and the network and promoting the development of the digital economy. However, in the current network architecture represented by TCP / IP, key mechanisms such as parsing discovery, network awareness, and routing computation have some drawbacks, making it difficult for them to serve as the foundation for new applications that integrate communication and computing.

[0053] By adopting an open internet routing architecture, the existing host interconnection design is systematically transformed, providing comprehensive, open, and guaranteed intrinsic service interconnection support for emerging application scenarios such as computing-network convergence. Based on this routing architecture, user terminals are enabled to initiate location-independent, efficient end-to-end connections according to service type, providing an effective way for the network to proactively acquire computing power and bandwidth requirements. The two-level, two-layer routing not only achieves routing orchestration of both computing and network resources but also ensures the feasibility and effectiveness of deploying this routing architecture under different levels of network openness.

[0054] However, due to the different network facilities, architectures, designs, and solutions of service providers, and the data security and privacy issues they face, it is currently difficult for service providers to interconnect and share their specific services. Furthermore, most new network technologies and solutions require extensive modifications to existing network infrastructure, making them difficult to deploy and implement.

[0055] In view of this, the technical solution of this invention provides a routing architecture for an easily deployable open interconnected network of computing power services. In the two-level routing, the first-level routing orchestrates routes from network access devices to network output devices, and the second-level routing orchestrates routes from network output devices to target service-providing nodes. Together, they provide complete end-to-end routing for service requests issued by user terminals. In the two-layer routing, the service layer routing completes routing orchestration while considering computing resources, and the "service layer + infrastructure layer" collaborative routing completes hop-by-hop routing orchestration by jointly considering computing and network resources. Mechanistically, the two-level routing reuses existing server equipment, protecting the autonomy and privacy of the service provider's internal network while reducing the pressure on the control plane and forwarding plane; the two-layer routing reuses existing routing and forwarding equipment and provides routing solutions for different levels of network openness; in terms of resource adaptation, the two-level routing achieves on-demand adaptation of computing resources, and the two-layer routing achieves on-demand adaptation of computing and network resource collaboration. The two complement each other, providing a feasible approach for the large-scale deployment of the two-layer, two-level routing architecture.

[0056] This embodiment provides a routing architecture for an open interconnection network for computing power services, such as... Figure 1 As shown, the architecture includes: user terminal 1, network access device 2, network management device 3, network control center 4, network output device 5, and server 6.

[0057] User terminal 1 is used to initiate service request messages and transmit service request messages to network access device 2.

[0058] User terminal 1 is the individual or device that ultimately uses the network service, such as a computer, smartphone, tablet, or other network device. Specifically, when a user initiates a service request on user terminal 1, such as accessing a webpage, downloading a file, or sending an email, the request is encapsulated into a data packet (service request message) and then enters the network through network access device 2.

[0059] Network access device 2 is used to receive service request messages sent by user terminal 1 and to forward the service request messages to network management device 3.

[0060] Network access device 2, acting as the ingress gateway, receives service request messages from user terminal 1 and forwards them to network management device 3 for processing. Network management device 3 parses, analyzes, and processes these requests, then makes corresponding decisions based on the current network status and policies, such as determining the optimal routing path and allocating resources.

[0061] The network control center 4 is used to notify the network management device 3 of the network resource status of each WAN routing node 7; wherein, the network resource status includes the network topology and resource status between each WAN routing node 7.

[0062] Network control center 4 is used to represent the WAN centralized controller and to announce the network resource status of each WAN routing node 7. WAN routing nodes 7 refer to routers connecting different areas or networks, responsible for forwarding data packets in the WAN and maintaining the WAN topology and network resource status. Network topology refers to the connection relationships and path information between various routing nodes in the WAN, including physical connections, logical connections, and the location relationships between routers. Resource status refers to the resource utilization of each routing node, such as bandwidth utilization, data packet forwarding speed, and latency. Specifically, when network control center 4 obtains the network resource status of each WAN routing node 7, it announces this status information to network management device 3. Network management device 3 uses this information to monitor the overall WAN operation status, including real-time network topology and resource utilization, in order to make corresponding network adjustments and optimizations.

[0063] Server 6 represents a service provider that offers various essential services to users, such as Internet Service Providers (ISPs), cloud service providers, and software service providers. Specifically, in a network environment, Server 6 runs multiple service-providing nodes, such as servers, virtual machines, and storage devices, which provide different network services.

[0064] Network output device 5 is used to receive the computing resource status of each service providing node sent by server 6, and to forward the computing resource status to network management device 3.

[0065] Network output device 5 represents the egress gateway. Located between the local area network (LAN) and the wide area network (WAN), it connects the internal network to the external internet. Network output device 5 is used for exchanging and forwarding data between the internal and external networks, forwarding data packets from the internal network to the external network and vice versa.

[0066] Network output device 5 receives computing resource status information from various service nodes of server 6 through a connection established with server 6. Server 6 has multiple service providing nodes, each of which periodically or in real-time sends computing resource status information, including CPU utilization, memory usage, disk space utilization, network bandwidth usage, etc. Specifically, when network output device 5 receives computing resource status information from server 6, it transmits this information to network management device 3.

[0067] Network management device 3 is used to characterize the intelligent control system and to parse service request messages to obtain parsing results. When network management device 3 senses the operating status of the infrastructure layer, it is also used to generate a first orchestration result between the first network output device and the target service provider node based on the parsing result, computing resource status, and network resource status. The first orchestration result includes hop-by-hop routing information corresponding to the service request message, which is generated based on the network resource status. Network management device 3 is also used to distribute the first orchestration result to network access device 2. Network access device 2 is also used to send the service request message to WAN routing node 7 so that the service request message is sent to the first network output device according to the specified path and strategy in the first orchestration result. The first network output device is used to send the service request message to the target service provider node according to a preset routing strategy.

[0068] Network management device 3 receives a service request message and parses it to obtain relevant information and parameters. For example, network management device 3 needs to parse the message format, such as identifying the header, body, and trailer, as well as the separators or flags between different parts. It extracts the values ​​of various fields from the message, such as the target address, request type, and parameters; these fields are usually identified in specific locations or markers within the message. Semantic understanding of the extracted fields is required, that is, understanding the meaning and function of the fields. For example, parsing the target address might involve determining which service node or server the request should be routed to. Error handling is also necessary, including detecting and handling format errors, missing fields, or invalid content in the message.

[0069] Network management device 3 monitors the operational status of the infrastructure layer, which includes the network control center 4 and various WAN routing nodes 7.

[0070] When the network management device 3 is able to perceive the operating status of the infrastructure layer, based on the results obtained from parsing the service request message, as well as the obtained computing resource status and network resource status, the network management device 3 generates the first orchestration result. This result is generated based on the current computing resource status and network resource status, taking into account factors such as the load and topology of each device in the network.

[0071] The first arrangement result includes hop-by-hop routing information corresponding to the service request message. Hop-by-hop routing information refers to the transmission path of the service request in the network, that is, the route path from the starting point to the target service providing node. This routing information is generated based on the current network resource status to ensure that the request can be transmitted along the optimal path. That is, network management device 3 considers factors such as the load of each device in the network, link status, and topology, and selects the best path to transmit the service request in order to improve network performance and service quality.

[0072] Based on the first orchestration results generated above, network management device 3 sends these results to network access device 2. These orchestration results include hop-by-hop routing information and transmission strategies for service request packets, guiding how service requests are transmitted in the network. Upon receiving the first orchestration results from network management device 3, network access device 2 is responsible for actually sending the service request packets to WAN routing node 7. WAN routing node 7 acts as the exit point for service requests entering the WAN. Network access device 2 transmits the service request packets to the first network output device according to the hop-by-hop routing information and transmission strategies specified in the first orchestration results. These paths and strategies are determined based on the aforementioned network resource status and the calculation results of network management device 3, aiming to optimize the transmission efficiency of service requests and network performance. The first network output device is the network output device selected based on the first orchestration results.

[0073] The preset routing policy is a set of guidelines pre-configured by the administrator to determine the transmission path of service request packets in the network. Specifically, the first network output device makes routing decisions for service request packets according to the preset routing policy. Once the transmission path is determined, the first network output device sends the service request packet to the target service provider node. The target service provider node can be a specific server, application, cloud service, etc., used to process service requests and provide corresponding services.

[0074] When the network management device 3 is unable to perceive the operating status of the infrastructure layer, the network management device 3 generates a second orchestration result between the network access device 2 and the second network output device based on the parsing result and the computing resource status; the network management device 3 is also used to send the second orchestration result to the network access device 2; the network access device 2 is also used to send the service request message to the wide area network routing node 7 so that the service request message is sent to the second network output device according to the specified path and strategy in the second orchestration result.

[0075] The inability to perceive the operational status of the infrastructure layer is used to characterize processes handled only through the service layer. Specifically, the service layer includes: user terminal 1, network access device 2, network management device 3, network output device 5, and server 6.

[0076] When network management device 3 is unable to perceive the operational status of the infrastructure layer, it uses the parsing results and computing resource status to adjust the connection layout or configuration between network access device 2 and the second network output device. This process may involve rearranging connection paths, adjusting data traffic allocation, and optimizing network resource utilization to ensure network operating efficiency and performance. The second network output device is selected based on the second orchestration result. The second network output device may be the same as or different from the first network output device.

[0077] Network management device 3 generates a second orchestration result, which is a network configuration scheme generated based on the parsing results and the computing resource status. This result needs to be applied to the network so that the network operates according to the designed layout and strategy. Therefore, network management device 3 is responsible for sending this result to network access device 2.

[0078] Network access device 2 receives the second orchestration result from network management device 3 and processes the service request message according to the path and policy specified therein. Network access device 2 sends the service request message to WAN routing node 7 so that it can be forwarded to the second network output device according to the path and policy specified in the second orchestration result.

[0079] The routing architecture of the open interconnection network for computing power services provided in this embodiment of the invention requires the intelligent management module to have comprehensive management capabilities through "service layer + infrastructure layer collaboration" routing. However, it also supports more refined, flexible, and reliable end-to-end routing control, achieving full-process protection of transmission and computing and optimal utilization of network and computing resources. In other words, the system can allocate resources more accurately and improve the utilization rate of network and computing resources. By parsing service request packets through the network management device and generating hop-by-hop routing information based on network resource status, the system can effectively manage the network topology and resource status between the WAN routing nodes. This helps optimize the allocation and use of network resources and solve the interconnection problem between different network facilities and architectures. Through the functions of the network management device, the system can generate and distribute preset routing policies to securely send service request packets to the target service provider node. This secure routing policy helps service providers ensure the security and integrity of data during transmission, thereby solving data security issues. Through the optimized routing scheme and resource management of this system, existing infrastructure can be utilized more effectively, reducing the transformation costs brought by new technologies and ensuring smooth interconnection, sharing, and deployment of services. It enables smooth network evolution based on existing network infrastructure, meaning the system can be upgraded gradually without impacting existing network facilities, ensuring service continuity and stability. Users only need to provide specific requirements, eliminating the need to select service providers, which simplifies the user access process, improves user experience, and reduces operational difficulty. The two-layer routing mechanism of service layer routing and "service layer + infrastructure layer collaborative" routing each has its advantages and disadvantages, but they complement each other, providing a feasible approach for large-scale network deployment and optimizing large-scale network deployment through complementary methods.

[0080] In some alternative implementations, the routing architecture of the above-mentioned open interconnection network for computing power services can be referred to as a two-level, two-layer routing architecture.

[0081] The two-level, two-layer routing architecture includes two levels of routing: Level 1 routing, used to orchestrate routes from network access device 2 to network output device 5; and Level 2 routing, used to orchestrate routes from network output device 5 to the service provider node. The Level 1 and Level 2 routes combine to provide complete end-to-end routing for service requests.

[0082] The two-level, two-layer routing architecture includes two routing layers: a service layer, which treats any node between user terminal 1 and server 6 as a directly connected node if a reachable path exists, thus masking the actual topology of the infrastructure layer; and a service layer, which also performs route orchestration while considering the state of computing resources. The service layer and the infrastructure layer work together to perform hop-by-hop route orchestration by jointly considering the states of computing and network resources.

[0083] Two-level routing is used to provide the optimal selection of computing power service resources based on the network load status and real-time resource status. It can restore all service request packets carrying SIDs to ordinary IP packets, so that they can be forwarded without obstacles through existing LAN devices.

[0084] Specifically, such as Figure 2 As shown, in a first-level routing scenario, user terminal 1 (the entry gateway) periodically interacts with network management device 3 (the intelligent control system) to achieve coarse-grained status updates of the entire network's computing power. Network management device 3 intelligently plans the tasks of numerous network access devices 2 based on the real-time computing power load of the entire network, according to the service request packets from network access devices 2. After receiving and deploying the routing entries issued by network management device 3, network access devices 2 encapsulate a new IP header for each service request packet based on the IP address of the destination network output device 5, and then hand it over to WAN routing node 7 for forwarding according to the routing policy specified by server 6, ultimately reaching network output device 5, completing the first-level routing process.

[0085] Specifically, such as Figure 3 As shown, in a two-level routing scenario, network output device 5 is used to detect the service-providing nodes of fixed server 6 in real time, enabling fine-grained status updates of computing power within the server's private domain network. Based on the task requests transmitted from network access device 2, network output device 5 performs intelligent planning and scheduling according to the computing power load within the private domain network. Since there is generally a pre-set load balancing strategy within the server's local area network, network output device 5 can select the best node from the set of nodes that can provide the service corresponding to the SID as the destination of the service request, based on this strategy and the requirements of the service request. Simultaneously, it replaces the SID and destination port number fields in the original service request packet with the IP address and service port number of the corresponding node recorded in the service list. Then, it forwards the packet (with the outer header removed from network access device 2) within the local area network to complete the delivery.

[0086] The routing architecture of the open interconnection network for computing power services provided in this embodiment of the invention has two levels of routing. The first level of routing orchestrates routes from network access devices to network output devices, while the second level orchestrates routes from network output devices to target service-providing nodes. Together, they provide complete end-to-end routing for service requests. In this two-layer routing, the service layer routing completes routing orchestration while considering computing resources, while the "service layer + infrastructure layer" collaborative routing completes hop-by-hop routing orchestration by jointly considering computing and network resources. Mechanistically, the two-level routing reuses existing server equipment, protecting the autonomy and privacy of the service provider's internal network while reducing pressure on the control plane and forwarding plane. It also reuses existing routing and forwarding equipment and provides routing solutions for different levels of network openness. In terms of resource adaptation, the two-level routing achieves on-demand adaptation of computing resources, and the two-layer routing achieves on-demand adaptation of computing and network resource collaboration. These two aspects complement each other, providing a feasible approach for the large-scale deployment of the two-level, two-layer routing architecture.

[0087] According to an embodiment of the present invention, a routing method embodiment for an open interconnection network for computing power services is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0088] This embodiment provides a routing method for an open interconnection network of computing power services. Figure 4 This is a flowchart of a routing method for an open interconnection network for computing power services according to an embodiment of the present invention, such as... Figure 4 As shown, the process includes the following steps:

[0089] In step S101, the user terminal initiates a service request message and reports the service request message to the network management device through the network access device.

[0090] In a network environment, user terminal 1 may need to access a service, such as browsing web pages, downloading files, or sending emails. When a user needs to access a service, they initiate a service request. This request can be triggered by inputting relevant information through a user interface (such as a browser or application), or it can be initiated automatically by the system, such as a request triggered by a scheduled task or system event. Once the user initiates a service request, the system generates a service request message based on the requested service. A service request message is a data packet containing information about the requested service, such as the service type, destination address, and data content. This message typically includes various parameters and identifiers for subsequent processing and routing. The generated service request message is reported to network management device 3 via network access device 2. Network access device 2 is a key node in the network, responsible for receiving requests from user terminal 1 and forwarding them to other nodes or services in the network. Transmission is usually performed using network protocols (such as TCP / IP) and can be completed through various communication methods (such as wired or wireless).

[0091] In step S102, the server sends the computing resource status of each service provider node to the network management device through the network output device.

[0092] A service provider node is a node or server in the network that performs a specific service function. Computational resource status refers to the current resource utilization and performance indicators of a service provider node, including its load, available computing resources, and processing capacity. Server 6 transmits the computational resource status of the service provider node to network output device 5, enabling network output device 5 to obtain the resource status of each service provider node in the network.

[0093] Network output device 5 transmits computing resource status to network management device 3, enabling more intelligent resource management and optimization. Network management device 3 can adjust network resource allocation and routing strategies in real time based on the received computing resource status information to maximize network performance, service quality, and resource utilization efficiency.

[0094] In step S103, the network control center notifies the network management device of the network resource status of each WAN routing node.

[0095] The network resource status includes the current resource utilization and performance indicators of WAN routing node 7. Network control center 4 transmits the network resource status of WAN routing node 7 to network management device 3, enabling network management device 3 to obtain the resource status of the entire WAN. When processing service requests, network management device 3 needs to consider the overall network resource status in order to make optimal routing and resource allocation decisions.

[0096] In step S104, the network management device parses the service request message to obtain the computing power requirement and network bandwidth requirement corresponding to the service request message.

[0097] Network management device 3 parses the received service request messages to determine the specific request content contained in the messages, such as the type of service the user needs to access, the expected service quality, and the duration. During the parsing process, network management device 3 obtains the computing power requirements and network bandwidth requirements from the messages. Computing power requirements refer to the amount of computing resources needed to execute the requested service, such as the CPU and GPU resources that may be needed to process data or run applications. Network bandwidth requirements refer to the data transmission rate required during service transmission, which directly relates to data transmission efficiency and latency. Based on the parsed computing power requirements and network bandwidth requirements, network management device 3 formulates corresponding resource allocation strategies, such as allocating sufficient computing resources to the service request and adjusting network configuration to meet bandwidth requirements, ensuring that the service can run according to the user's expected performance.

[0098] Computing power requirements refer to the computing resources needed for a request, such as service type requirements, computing power capacity requirements, latency guarantee requirements, and optimization target requirements. Service type requirements refer to the specific computing power service type corresponding to the service request message. Different services may require different types of computing resources to perform specific tasks or provide specific functions. Computing power capacity requirements refer to the computing power requirements of a specific computing power service, including the quantity and performance requirements of computing resources. Latency guarantee requirements refer to the completion time limit corresponding to a specific computing power service; some services have strict time requirements for completing tasks and must be completed within a specified time limit. Optimization target requirements refer to the comprehensive requirements of user terminal 1 for completing the computing power task, including maximizing schedulable traffic, minimizing total latency, and link load balancing. The computing power service task must be completed within the optimization target requirements; therefore, task data is transmitted according to the methods for meeting the optimization target requirements.

[0099] Bandwidth requirement refers to the network bandwidth needed for a request, which is used to transmit data at a certain speed and capacity.

[0100] Step S105: When the network management device cannot perceive the operating status of the infrastructure layer, the network management device generates a second orchestration result between the network access device and the second network output device based on the computing power requirements and computing resource status; the network management device sends the second orchestration result to the network access device; the network access device sends the service request message to the wide area network routing node so that the service request message can be sent to the second network output device according to the specified path and strategy in the second orchestration result.

[0101] When network management device 3 cannot obtain detailed status information of current infrastructure resources, it can generate a second orchestration result based on known computing power requirements (information contained in service request messages) and known computing resource status (computing resource status of each service-providing node sent by the server). The second orchestration result is a service request processing strategy formulated by network management device 3 based on currently available information, including the service request routing path, priority, and resource allocation strategy. After generating the second orchestration result, network management device 3 distributes this information to network access device 2. Network access device 2 acts as a bridge between network management device 3 and user terminal 1, responsible for the actual service request forwarding and routing.

[0102] After receiving a service request message from a user, network access device 2 will forward these service request messages to WAN routing node 7 according to the path and policy specified in the second orchestration result. WAN routing node 7 is responsible for guiding the transmission path of data packets throughout the network, ensuring that the data can reach the target device (second network output device) according to the predetermined route.

[0103] Specifically, step S105 includes: the network management device generating a second orchestration result between the network access device and the second network output device based on a preset service layer routing model, computing power requirements, and computing resource status.

[0104] When the network status of the infrastructure layer is not visible to network management device 3, network management device 3 only needs to complete the routing orchestration in the service layer view. Since the service layer is a virtual network layer built on top of the infrastructure layer, from the service layer perspective, as long as there is a reachable path between user terminal 1 and server 6, it can be regarded as a directly connected node, thus shielding the actual topology connection of the infrastructure layer. From this perspective, network management device 3 only needs to consider the computing power requirements of the service request and the computing resource status of server 6 for routing orchestration. The routing decision is also simplified to the optimization selection problem of the server's egress gateway. The routing decision model at this time can be called the service layer routing model.

[0105] Specifically, the computing power state matrix S is introduced. C The computing power status matrix S is used to describe the real-time computing resource usage status of each service provider node. C The definition is as follows:

[0106]

[0107] in, Indicates the egress gateway e m All service-providing nodes within the local area network of the server provide SIDs. n The sum of computing resources used by the corresponding services; f(S) CThis represents the evaluation metric for the primary routing orchestration results, using... Let the set of ingress gateways, the set of egress gateways, and the set of SIDs represent the ingress gateways, egress gateways, and SIDs, respectively. Then, the service layer routing orchestration problem can be modeled as:

[0108] max(min){f(S C )} (1)

[0109]

[0110]

[0111]

[0112]

[0113] Equation (1) represents a general optimization objective, such as achieving load balancing; Equation (2) indicates whether a service request corresponding to a request from network access device 2 should select network output device 5, which is an indicative function; Equation (3) constrains that a service request requesting the same type of service from the same network access device 2 can only select one destination network output device 5; Equation (4) calculates the sum of the computing power requirements of all corresponding services provided by server 6 where network output device 5 is located; Equation (5) constrains that the computing power requirements of each type of service provided by server 6 cannot exceed the upper limit of computing resources reserved for that service, which represents the upper limit of computing resources reserved by all nodes in the local area network of the server where network output device 5 is located for the corresponding service.

[0114] Based on the above model, network management device 3 can solve the service request orchestration optimization of computing resources by designing a scheduling algorithm. After receiving and deploying the routing entries issued by network management device 3, network access device 2 will encapsulate a new IP header for each service request packet according to the IP address of the destination egress gateway, and hand it over to WAN routing node 7 for forwarding according to the routing policy specified by the server, finally reaching network output device 5, completing the first-level routing process.

[0115] Step S106: When the network management device senses the operating status of the infrastructure layer, the network management device generates a first orchestration result between the first network output device and the target service provider node based on the computing power requirement, network bandwidth requirement, computing resource status, and network resource status; the network management device sends the first orchestration result to the network access device; the network access device sends the service request message to the wide area network routing node so that the service request message can be sent to the first network output device according to the specified path and strategy in the first orchestration result; the first network output device sends the service request message to the target service provider node according to the preset routing strategy.

[0116] Network management device 3 is capable of sensing the operational status of the infrastructure layer, i.e., acquiring information such as load and availability of network devices and other infrastructure. Based on the sensed infrastructure layer operational status, network management device 3 combines information such as computing power requirements, network bandwidth requirements, computing resource status, and network resource status to generate a first orchestration result. The first orchestration result includes information such as determining the processing path of service requests, the priority of resource allocation, and the target service provider node. After generating the first orchestration result, network management device 3 distributes this information to network access device 2. Network access device 2 sends these service request packets to WAN routing node 7 according to the path and policy specified in the first orchestration result. WAN routing node 7 transmits the service request packets to the first network output device according to the path and policy specified in the first orchestration result. The first network output device sends the service request packets to the target service provider node according to a preset routing policy.

[0117] Specifically, step S106 includes: the network management device generating a first orchestration result between the first network output device and the target service provider node based on a preset service layer and infrastructure layer collaborative routing model, computing power requirements, bandwidth requirements, computing resource status and network resource status.

[0118] Once network management device 3 possesses the ability to perceive the infrastructure layer, it can use a two-layer routing mechanism. This mechanism jointly considers the computing resource status of the service layer and the network resource status of the infrastructure layer, as well as the computing power and bandwidth requirements corresponding to the service requests, to orchestrate hop-by-hop routes for the service requests. This routing decision model can be called a "service layer + infrastructure layer" collaborative routing model.

[0119] Specifically, the bandwidth state matrix S is introduced. B The bandwidth status matrix S is used to characterize the real-time bandwidth resource status of each WAN routing node 7. B The definition is as follows:

[0120]

[0121] in, Indicates WAN routing node 7r m to r n The bandwidth occupied by a unidirectional link; if there is no direct link between the two nodes, the element value is 0; g(S C ,S B The ) represents the evaluation metric for the secondary routing orchestration results. Therefore, the collaborative routing orchestration problem between the service layer and the infrastructure layer can be modeled as follows:

[0122] max(min){g(S C ,S B (6)

[0123]

[0124]

[0125]

[0126]

[0127]

[0128] Equation (6) can represent a general optimization objective, such as achieving joint balancing of bandwidth and computing resources. Equation (7) indicates whether a service request corresponding to a request accessed from network access device 2 should select a path from the set of legal paths. Equation (8) represents the set of all legal paths from network access device 2 to network output device 5 that can provide the corresponding service. Equation (9) constrains that a service request requesting the same type of service from the same network access device 2 can only select one legal path. Equation (10) calculates the sum of the bandwidth requirements of service requests through each link of the infrastructure layer. Equation (11) constrains that the bandwidth requirements carried by each link of the infrastructure layer cannot exceed the upper limit of the bandwidth resources of that link. Equation (12) represents the upper limit of the bandwidth of a unidirectional link from node to node. Equation (13) indicates whether a path is selected is associated with whether the destination network output device 5 on the path is selected.

[0129] Specifically, based on the above model, network management device 3 can solve for the service request orchestration that jointly optimizes bandwidth and computing resources by designing a scheduling algorithm. In particular, the orchestration result of this model includes hop-by-hop routing information. Therefore, in addition to encapsulating a new IP header for each service request packet according to the routing entries issued by network management device 3, network access device 2 will also encapsulate forwarding path information into the SRH extension header of the data request packet to achieve hop-by-hop forwarding control. After reaching network output device 5, the data packet will undergo secondary routing to complete the final packet delivery.

[0130] In some alternative implementations, to support latency-sensitive services such as real-time cloud rendering and holographic communication, the future integrated computing and network architecture will also need to have latency protection capabilities.

[0131] Specifically, if WAN routing node 7 and the directly connected link have time deterministic guarantee capabilities, the service layer and infrastructure layer collaborative routing model can be extended to a service layer and infrastructure layer collaborative deterministic routing model to achieve determinism in the time dimension, and can then be modeled as follows:

[0132] max(min){g(S C ,S B (12)

[0133]

[0134]

[0135]

[0136]

[0137] Equation (12) represents a general optimization objective such as network resource balancing; Equation (13) calculates that the service delay is equal to the sum of the transmission delay and the computation delay. and Let represent the transmission delay of the service request corresponding to the SID accessed from the ingress gateway i and the calculation delay of the service request corresponding to the SID being received at the egress gateway e, respectively; Equation (14) calculates that the transmission delay is equal to the sum of the upper limit of the delay of all deterministic links on the path and the transmission delay threshold within the server's local area network. For link (r) m ,r n The deterministic delay upper limit of ) is θ, which is a preset threshold used to describe the upper limit of the delay of service requests transmitted within the server's local area network. This variable can be used to reserve delay for the forwarding process on nondeterministic links within the server's local area network; Equation (15) calculates that the computation delay is equal to the sum of the computational amounts of the corresponding services provided within the local area network divided by the maximum number of operations, l i,SID This represents the computational load of the service corresponding to the request SID accessed from the ingress gateway i; Equation (16) constrains the service latency to not exceed the upper limit of the latency requirement of the corresponding service request. This indicates the upper limit of latency requirements for service requests corresponding to the service SID accessed from the ingress gateway i.

[0138] A diagram of two-layer routing is shown below. Figure 5 As shown, this includes service layer routing and "service layer + infrastructure layer collaborative" routing. While two-layer routing requires network management device 3 to have comprehensive control capabilities, it supports more granular, flexible, and reliable end-to-end routing control. This allows for the expansion of complex traffic strategies, such as latency deterministic guarantee policies, ultimately achieving end-to-end protection of transmission and computation and optimal utilization of network and computing resources. Simultaneously, the deployment of hop-by-hop routing information via SRv6 avoids the practical deployment challenges associated with upgrading the protocol stack of existing routing devices to support SID resolution.

[0139] The routing method for open interconnected computing power services provided in this invention simplifies the user process by allowing users to provide specific requirements without needing to filter service providers. This reduces management and selection costs, enabling users to focus more on their needs and improving user experience. This routing method enables smooth network evolution based on existing network infrastructure, effectively integrating with existing infrastructure and reducing major modifications to the network infrastructure, thus lowering deployment and upgrade risks and costs. The two-layer routing mechanism—service layer routing and "service layer + infrastructure layer collaborative" routing—complements each other, supporting more granular, flexible, and reliable end-to-end routing control. It ensures routing stability and the timeliness of measurement information, maintaining routing stability while promptly acquiring and utilizing the latest network and computing resource statuses. The intelligent management module performs primary routing between network access devices and network output devices based on computing power requirements and computing resource status, and secondary routing between network output devices and target service provider nodes based on computing power requirements, network bandwidth requirements, computing resource status, and network resource status. This achieves full-process protection of transmission and computing and optimal utilization of network and computing resources. This comprehensive resource scheduling and routing control helps improve resource utilization and overall performance.

[0140] In some alternative implementations, such as Figure 6 As shown, the routing method of the above-mentioned open interconnection network for computing power services can be divided into an initialization phase, a control phase, and a service phase.

[0141] During the initialization phase, the service agent deployed by the server resource pool performs service registration, announcing a service list consisting of the SID, node IP address, and port number of the available services in the local resource pool to the connected network output device 5. Network output device 5 maintains the list of available services for all nodes in the server resource pool locally and simultaneously announces the SIDs of all services available in the server's local resource pool to network management device 3. Network management device 3 maintains information on all network output devices 5 within its control scope and their corresponding available services, and distributes a list containing all available service SIDs to network access device 2 and user terminal 1. When service migration occurs, both the old and new service providers need to update their available service information with network management device 3.

[0142] During the control phase, the service provider node periodically notifies the network output device 5 of its computing resource usage and the lifecycle health status of the service instance. Since computing resource usage is typically real-time, to prevent this from affecting routing convergence, the network output device 5 performs secondary statistical processing on the collected fast-changing computing resources, generating slow-changing computing resources at longer intervals and notifying the intelligent management system. The network management device 3 also receives network resource notifications from the network control center 4, including information such as topology connections, bandwidth usage, and packet loss rate. Simultaneously, the network access device 2 statistically analyzes the computing power and bandwidth requirements of all service requests generated by user terminals 1 within the local area network and periodically notifies the network management device 3 of these computing network requirements. Upon receiving resource usage and requirement information, the network management device 3 performs routing orchestration as needed, generating routing entries indexed by SIDs for each network access device 2 and distributing these routing entries to the network access devices 2.

[0143] During the service phase, when initiating a service request, user terminal 1 encapsulates the service request content into a message and generates a corresponding SID locally based on the type of service required. This SID is then filled into the destination IP field of the message to complete the encapsulation. The encapsulated original message is forwarded by user terminal 1 to network access device 2. Upon receiving the message, network access device 2 first parses the SID in the message and uses the SID as an index to query its local routing table. Then, it encapsulates an IP header outside the original message. The source IP in the header is the IP address of network access device 2, and the destination IP is the IP address of the destination egress gateway in the matched routing table entry. If the routing table entry contains hop-by-hop forwarding path information, this information will be displayed in SRv6 format. The IP packet is encapsulated in an SRH extension header. The encapsulated IP packet is then sent to WAN routing node 7 for forwarding, ultimately reaching the corresponding network output device 5. Upon receiving the service request packet, network output device 5 removes the outer IP header and parses the SID in the packet. Then, based on the server's LAN routing policy, it selects the specific service provider node, replaces the SID and destination port number fields in the original service request packet with the node's IP address and the port number corresponding to the service, and forwards the packet to the service provider node to complete the delivery. Simultaneously, network output device 5 also generates a routing table entry locally. Subsequent packets from the same service request can directly match the routing table based on the source IP and SID to be delivered to the same service provider node.

[0144] The routing method for the open interconnection network of computing power services provided in this invention enables user terminals to initiate location- and home-independent service connections by introducing SIDs. This achieves integrated service awareness and service provision from a network perspective, providing an architectural foundation for global service orchestration and optimization schemes for common computing services. Simultaneously, based on current host / network segment routing and SRv6 routing mechanisms, it avoids the overhead of adding service identifier resolution functionality to existing network infrastructure, promoting a smooth evolution of the network from traditional host interconnection to service interconnection. In the future, with the widespread application of the routing architecture of the open interconnection network of computing power services, end-side devices and switching devices will gradually be updated to support IP / SID dual-stack support. At that time, the open interconnection network of computing power services will achieve further functional optimization and enhancement in areas such as distributed control, mobility support, and routing table aggregation.

[0145] This embodiment provides a routing method for an open interconnection network of computing power services. Figure 7 This is a flowchart of a routing method for an open interconnection network for computing power services according to an embodiment of the present invention, such as... Figure 7 As shown, when the network management device senses the operational status of the infrastructure layer, the process includes the following steps:

[0146] In step S201, the user terminal generates a corresponding identifier locally based on the type of the service request content, and encapsulates the identifier into the service request message to obtain the first service request message.

[0147] The identifier is used to characterize the Service Identifier (SID), which uniquely identifies a specific service or resource within the network. The identifier is used to uniformly describe the various services within the routing architecture of the open interconnected network for accessing computing power services. These services may include, but are not limited to, network connectivity, security, and data transmission.

[0148] The identifier is used to support efficient end-to-end communication between user terminal 1 and the service provider node, enabling centralized management and efficient requesting of computing resources. User terminal 1 can generate a SID locally to initiate communication without needing to send an address resolution request to a centralized server through the traditional DNS mechanism, thus decoupling service resources from location. Specifically, user terminal 1 generates or obtains the corresponding SID locally based on the service type and requirements of the service request message, and directly uses the SID to initiate a communication connection request. User terminal 1 encapsulates the content of the service request into a message and generates a corresponding SID locally based on the type of service required. This SID is filled into the destination IP field of the message to complete the message encapsulation, forming the first service request message. That is, the first service request message is a message in which the identifier is encapsulated within the original service request message and forwarded by user terminal 1 to network access device 2.

[0149] In step S202, after receiving the first service request message, the network access device uses the identifier as an index to query the locally deployed routing entries, and encapsulates the corresponding routing information into the first service request message to obtain the second service request message; and reports the second service request message to the network management device.

[0150] Network access device 2 uses an identifier as an index to query locally deployed routing entries and encapsulates the corresponding routing information into a service request message to obtain a second service request message. Specifically, network access device 2 uses the identifier provided by user terminal 1 as an index to look up entries in its locally stored routing table. That is, the identifier is used to find the corresponding routing information in the routing table to determine how to handle this specific service request. When network access device 2 finds a routing entry corresponding to the identifier, it extracts the corresponding routing information from the routing table and encapsulates this information into the service request message to ensure that the service request is correctly routed and processed in the network. In other words, the second service request message is a message with routing information encapsulated within the first service request message.

[0151] During the packet encapsulation process, network access device 2 can not only encapsulate the original packet with a new outer IPv6 packet header, but also explicitly identify the hop-by-hop forwarding path through the SRH extension header to achieve traffic engineering, thus avoiding the protocol stack upgrade required by existing relay routing devices that provide traffic forwarding in the wide area network to resolve service request packets.

[0152] In step S203, the network management device parses the second service request message to obtain the computing power requirement and network bandwidth requirement corresponding to the service request message; and generates a first orchestration result between the first network output device and the target service provider node based on the computing power requirement, network bandwidth requirement, computing resource status and network resource status; and sends the first orchestration result to the network access device.

[0153] Network management device 3 is responsible for the unified management of network devices and services within its management domain. In other words, network management device 3 can monitor, configure and manage various network devices (such as routers, switches, firewalls, etc.) and the various network services they provide (such as routing, security services, load balancing, etc.).

[0154] Network management device 3 receives and maintains the computing resource status within each server resource pool and the network resource status of the WAN routing nodes 7, which are periodically probed by network control center 4. It also periodically collects bandwidth and computing power demand information for each network access device 2 at the service level. Based on these two types of information, network management device 3 can coordinate network and computing resources to orchestrate the optimal forwarding path for access service requests on demand, and then distribute the first orchestration result to network access device 2 through the control channel.

[0155] In step S204, the network access device sends the second service request message to the wide area network routing node according to the first arrangement result for routing forwarding.

[0156] Network access device 2 processes the second service request based on the first orchestration result generated above. That is, network access device 2 will refer to the path and policy specified in the first orchestration result to determine how to process the second service request. According to the instructions of the first orchestration result, network access device 2 sends the second service request message to WAN routing node 7, which is located at the edge of the network and is responsible for managing the connection between the WAN and the local network.

[0157] In step S205, the WAN routing node sends the second service request message to the first network output device according to the specified path and strategy in the first orchestration result.

[0158] WAN routing node 7 examines the previously generated first orchestration result, which specifies the processing path and strategy for the second service request. This may include, for example, the address of the destination location and instructions on how to send the request to that location. Based on the instructions in the first orchestration result, WAN routing node 7 sends the second service request message to the first network output device, for example, traversing the request through a series of network nodes and links to ensure it reaches the destination node. The first network output device is an interface of WAN routing node 7 responsible for transmitting data to the network where the target service provider node resides. Therefore, WAN routing node 7 transmits the second service request message to the first network output device for subsequent routing and processing.

[0159] Specifically, WAN routing node 7 performs host routing or network segment routing based on the destination IP address according to the routing policy inside server 6. The routing policy inside server 6 refers to the routing rules or policies configured on the server, used to determine how to send data packets from the server to the target device. These routing policies can make decisions based on different conditions, such as destination IP address, source IP address, service type, etc. Host routing refers to routing data packets directly to the target host (a single device). When WAN routing node 7 performs host routing, it determines which specific host the data packet should be sent to based on the destination IP address. Network segment routing refers to routing data packets to a group of hosts in the target network. When WAN routing node 7 performs network segment routing, it sends the data packet to all hosts in the target network based on the destination IP address and network mask, etc. In this case, WAN routing node 7 selects to perform host routing or network segment routing based on the destination IP address in the data packet. According to the routing policy inside server 6, WAN routing node 7 checks the destination IP address and decides which routing method to use based on the configured rules.

[0160] When the control capabilities of WAN routing node 7 are opened to network management device 3, network management device 3 can add traffic policy scheduling based on routing policies, such as bandwidth guarantee policies and time slot-based delay deterministic guarantee policies. The traffic policy scheduling results are then deployed to WAN routing node 7 through network control center 4 as an intermediary. Specifically, when the control capabilities of WAN routing node 7 are opened to network management device 3, network control center 4 periodically collects the network status of WAN routing node 7, i.e., network resource usage, and notifies network management device 3 of the network status through the inter-controller information channel. Network management device 3 performs hop-by-hop path orchestration, and WAN routing node 7 can forward messages according to the path specified in the SRH extension header of the service request message. In addition, if the WAN routing node 7 also has specific traffic scheduling strategies such as bandwidth guarantee strategy and time slot-based delay deterministic guarantee strategy, and the infrastructure layer management and control capabilities are open to the network management device 3, the network management device 3 can add traffic strategy scheduling on the basis of the routing strategy, such as bandwidth guarantee strategy and time slot-based delay deterministic guarantee strategy, and deploy the scheduling results to the WAN routing node 7 through the network control center 4.

[0161] In some alternative implementations, the WAN routing node 7 can perform host routing or network segment routing based on the destination IP address according to the routing policy inside the server 6, or it can forward the message according to the path specified in the SRv6 header of the service request message.

[0162] Step S206: The first network output device parses the second service request message to obtain an identifier; selects a target service provider node based on the identifier; replaces the original Internet Protocol address and the original port number of the corresponding application in the second service request message with the target Internet Protocol address and the target port number of the target service provider node; and transmits the replaced second service request message to the target service provider node.

[0163] The first network output device parses the received second service request message to obtain the information contained therein. During this process, the first network output device extracts an identifier used to determine the target service type or target service provider node. Based on the parsed identifier, the first network output device selects an appropriate target service provider node to process the request according to pre-configured routing information or policies, ensuring that the request is delivered to the correct service node to obtain the corresponding service.

[0164] Before further processing the second service request message, the first network output device replaces the original Internet Protocol address and the original port number of the application to correctly redirect the request to the target service provider node, as the target service provider node may have a different network address and port configuration. Finally, the first network output device transmits the target service request message with the replaced address and port to the selected target service provider node. This ensures that the request accurately reaches the target node, thereby achieving request processing and service response, i.e., secondary routing.

[0165] The routing method for open interconnected computing power services provided in this invention employs a two-level routing approach to select the optimal computing power service resources based on network load and real-time resource status. It can restore all service request packets carrying SIDs to ordinary IP packets, enabling seamless forwarding through existing LAN devices. The two-layer routing simplifies the routing process, providing more refined, flexible, and reliable end-to-end routing control, ensuring the entire transmission and computation process and optimizing the utilization of network and computing resources. The two-layer routing jointly considers the computing resources of the service layer, the network resources of the infrastructure layer, and the computing power and bandwidth requirements corresponding to the service request, orchestrating hop-by-hop routes for the service request. User terminals generate identifiers locally and encapsulate them in service request packets. Network access devices can directly use these identifiers to query locally deployed routing entries, thereby reducing reliance on network management devices and improving the efficiency and response speed of route resolution. After sensing the operational status of the infrastructure layer, the network management device generates dynamic orchestration results based on the type of service request, computing power requirements, network bandwidth requirements, and real-time computing and network resource status. This enables intelligent routing and dynamic adjustment of service requests, improving resource utilization and network performance. The identifier generated by the user terminal remains unchanged throughout the service request process. Using this identifier to select the target service provider node enables transparent processing of service requests, improving network manageability and maintainability. The routing information encapsulated in the first service request message can be customized according to the type of service request content, allowing the network to better adapt to different types of service requests and provide more personalized services. After the first network output device parses the service request message and replaces the original Internet Protocol address and port number, secure processing of the service request is achieved, ensuring the legitimacy and security of the service request and preventing malicious attacks and unauthorized access.

[0166] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A routing architecture for an open interconnection network for computing power services, characterized in that, include: A network access device is used to receive service request messages sent by user terminals and to forward the service request messages to a network management device. A network output device is used to receive the computing resource status of each service providing node sent by the server, and to forward the computing resource status to the network management device. The network control center is used to notify the network management device of the network resource status of each WAN routing node; wherein, the network resource status includes the network topology and resource status among the various WAN routing nodes; The network management device is configured to parse the service request message to obtain a parsing result. When the network management device senses the operating status of the infrastructure layer, it is further configured to generate a first orchestration result between the first network output device and the target service provider node based on the parsing result, the computing resource status, and the network resource status. The first orchestration result includes hop-by-hop routing information corresponding to the service request message, which is generated based on the network resource status. The network management device is further configured to distribute the first orchestration result to the network access device. The network access device is further configured to send the service request message to the WAN routing node, so that the service request message is sent to the first network output device according to the specified path and strategy in the first orchestration result. The first network output device is configured to send the service request message to the target service provider node according to a preset routing strategy.

2. The architecture according to claim 1, characterized in that, Also includes: When the network management device is unable to perceive the operating status of the infrastructure layer, the network management device generates a second orchestration result between the network access device and the second network output device based on the parsing result and the computing resource status. The network management device is also used to send the second orchestration result to the network access device; The network access device is further configured to send the service request message to the wide area network routing node, so that the service request message is sent to the second network output device according to the specified path and strategy in the second orchestration result.

3. A routing method for an open interconnected network of computing power services, characterized in that, The method, applied to the routing architecture of the open interconnection network for computing power services as described in claim 2, comprises: The user terminal initiates a service request message and reports the service request message to the network management device through the network access device; The server sends the computing resource status of each service provider node to the network management device through the network output device; The network control center will notify the network management device of the network resource status of each WAN routing node; The network management device parses the service request message to obtain the computing power requirement and network bandwidth requirement corresponding to the service request message; When the network management device is unable to perceive the operating status of the infrastructure layer, the network management device generates a second orchestration result between the network access device and the second network output device based on the computing power requirements and the computing resource status; the network management device sends the second orchestration result to the network access device; the network access device sends the service request message to the wide area network routing node so that the service request message can be sent to the second network output device according to the specified path and strategy in the second orchestration result; When the network management device senses the operating status of the infrastructure layer, it generates a first orchestration result between the first network output device and the target service provider node based on the computing power requirement, the network bandwidth requirement, the computing resource status, and the network resource status. The network management device then sends the first orchestration result to the network access device. The network access device sends the service request message to the WAN routing node so that the service request message can be sent to the first network output device according to the specified path and strategy in the first orchestration result. The first network output device then sends the service request message to the target service provider node according to a preset routing strategy.

4. The method according to claim 3, characterized in that, When the network management device senses the operating status of the infrastructure layer, the method further includes: The user terminal generates a corresponding identifier locally based on the type of the service request content, and encapsulates the identifier into the service request message to obtain the first service request message; After receiving the first service request message, the network access device uses the identifier as an index to query the locally deployed routing entries, and encapsulates the corresponding routing information into the first service request message to obtain the second service request message; the second service request message is then reported to the network management device. The network management device parses the second service request message to obtain the computing power requirement and network bandwidth requirement corresponding to the service request message; and generates the first orchestration result between the first network output device and the target service provider node based on the computing power requirement, the network bandwidth requirement, the computing resource status and the network resource status; and sends the first orchestration result to the network access device. The network access device sends the second service request message to the wide area network routing node according to the first orchestration result for routing forwarding; The WAN routing node sends the second service request message to the first network output device according to the specified path and strategy in the first orchestration result; The first network output device parses the second service request message to obtain the identifier; selects the target service provider node according to the identifier, replaces the original Internet Protocol address and the original port number of the corresponding application in the second service request message with the target Internet Protocol address and the target port number of the target service provider node, and transmits the replaced second service request message to the target service provider node.

5. The method according to claim 3, characterized in that, The network management device is used to generate a second orchestration result between the network access device and the second network output device based on the computing power requirements and the computing resource status, including: The network management device generates a second orchestration result between the network access device and the second network output device based on a preset service layer routing model, the computing power requirement, and the computing resource status.

6. The method according to claim 3, characterized in that, The network management device generates a first orchestration result between the first network output device and the target service provider node based on the computing power requirement, the network bandwidth requirement, the computing resource status, and the network resource status, including: The network management device generates a first orchestration result between the first network output device and the target service provider node based on a preset service layer and infrastructure layer collaborative routing model, the computing power requirement, the bandwidth requirement, the computing resource status, and the network resource status.

7. The method according to any one of claims 3 to 6, characterized in that, The computing power requirements include service type requirements, computing power capacity requirements, and latency guarantee requirements. The service type requirement is the service type of the specific computing power service corresponding to the service request message. The computing power requirement refers to the size of the computing power requirement for the specific computing power service. The latency guarantee requirement refers to the completion time limit corresponding to the specific computing power service.

8. The method according to claim 5, characterized in that, The service layer routing model is represented as follows: ； in, , It indicates the real-time computing resource status of each service provider node; SID represents the various services in the routing architecture of the open interconnection network for accessing computing power services; Indicates the outbound gateway All service-providing nodes within the local area network of the server are provided with The sum of computing resources used by the corresponding services; This represents the evaluation index for the second arrangement result, using... These represent the ingress gateway set, egress gateway set, and SID set, respectively.

9. The method according to claim 8, characterized in that, The service layer and infrastructure layer collaborative routing model is represented as follows: ； in, , This indicates the real-time bandwidth resource status of each WAN routing node; Represents a WAN routing node arrive The bandwidth occupied by a one-way link; if there is no direct link between the two nodes, the element value is 0. This represents the evaluation index for the first arrangement result; Indicates from the ingress gateway Access request Whether the service request for the corresponding service selects the centralized path of the valid path; Indicates from the ingress gateway Available The complete set of all legal paths for the corresponding service's egress gateway; It is an indicator function; For a set of links; Indicates by the ingress gateway Access and request The sum of bandwidth requirements for all service requests. Represents a node To the node The bandwidth limit of a unidirectional link; This is an indicator function.

10. The method according to claim 9, characterized in that, If the WAN routing node and the direct link have time deterministic guarantee capabilities, the service layer and infrastructure layer collaborative routing model is extended to a service layer and infrastructure layer collaborative deterministic routing model to achieve determinism in the time dimension. The collaborative deterministic routing model between the service layer and the infrastructure layer is represented as follows: ； in, Represents the set of outbound gateways; Indicates the outbound gateway; Indicates service delay; and These represent the gateways from the ingress gateway. Access request The transmission latency of the service request for the corresponding service and at the egress gateway accept The computation latency of the service request corresponding to the service; For link The upper limit of deterministic delay, A preset threshold is used to describe the upper bound of the latency of service requests being transmitted within the service provider's local area network. This variable is used to reserve latency for the forwarding process on nondeterministic links within the service provider's local area network. It is an indicator function; Indicates from the ingress gateway Access request The computational load of the corresponding service; Indicates the outbound gateway All nodes within the service provider's local area network are The upper limit of computing resources reserved for the corresponding service; Indicates from the ingress gateway Access request The upper limit of latency requirements for service requests for the corresponding service.

Images