A computing power routing method for a white box switch

Abstract

The application discloses a computing power routing method for a white box switch. The application aims at the basic function of a network switch, adds a computing power routing module, and realizes a routing mechanism based on computing power optimization. First, the state of a network card is listened to, and computing power routing information captured by a network chip is written into a switch database, the database publishes relevant information to a service process, the service process recalculates routing after obtaining the computing power routing information from the database, and updates routing table items, so that computing power routing is realized. By introducing computing power routing, the computing resources of each node in the network can be maximally utilized, and the resource utilization rate is improved. The optimization capability enables the network to more efficiently utilize limited resources under the premise of ensuring performance, reduces network operation cost, and provides important support for distributed computing and edge computing.

Description

Technical Field

[0001] This invention relates to a computing power routing method for white-box switches, belonging to the field of data communication technology. Background Technology

[0002] In recent years, with the development of artificial intelligence (AI), connected vehicles, and the Internet of Things (IoT), edge computing and distributed computing have become important trends. These technologies require networks not only to transmit data efficiently but also to process data rapidly near the point of origin. To meet these requirements, traditional methods of network design and management are no longer sufficient. Traditional network routing algorithms mainly make decisions based on factors such as path length, bandwidth, and network congestion, while neglecting the crucial dimension of the computing power of network nodes.

[0003] However, in the current technological environment, computing power has become an indispensable factor, directly affecting the speed and efficiency of data processing, and consequently, the performance of the entire network system. Therefore, incorporating computing power as a component of routing calculation rules is particularly important. This means that the network must not only make routing decisions based on the optimal path for data transmission, but also consider the computing resource status of each node along the path to ensure that data can be processed quickly and effectively. In this way, the network can achieve intelligent allocation and scheduling of computing resources, ensuring that the performance requirements of critical applications are met, while improving the overall network resource utilization efficiency and service quality. Summary of the Invention

[0004] This invention aims to provide a computing power routing method for white-box switches. By monitoring the status of the network interface card (NIC) and writing the computing power routing information captured by the network chip into the switch database, the database publishes the relevant information to the service process. After obtaining the computing power routing information from the core database, the service process recalculates the route and updates the routing table entries, thereby realizing computing power routing.

[0005] The technical solution for achieving the technical objective of this invention is as follows:

[0006] A computing power routing method for white-box switches involves monitoring the status of network interface cards (NICs) and writing the computing power routing information captured by the network chip into the switch's database. The database then publishes the relevant information to the service processes. After obtaining the computing power routing information from the database, the service processes recalculate the routes and update the routing table entries, thereby achieving computing power routing.

[0007] The computing power routing method for white-box switches includes: S100, each functional module of the switch runs in an independent container, adopting a microservice architecture to reduce the coupling between business modules and facilitate functional upgrades and secondary development;

[0008] The S200 database, as a core component, stores the status information of the switch and serves as a middleware for message passing, running independently in a container. Information and data transmission between service processes are carried out through the database. The producer process writes data into the database, and then the database notifies the consumer process listening for the event. The consumer process retrieves the necessary information and data from the database.

[0009] S300, within the network operating system, runs a switching state service container, which undertakes network processing tasks;

[0010] The calculation of S400 and computing power routing is performed in the border gateway protocol container, and computing power routing is based on the extended implementation of the border gateway protocol.

[0011] S200 includes the following:

[0012] S210. The database runs in a database container. The database is divided into application database, configuration database, status database, network card data, and counter database.

[0013] S220, the application database stores the state generated by all application containers, including computing power information, routing information, next hop, and neighbors; it is the southbound entry point for all applications that wish to interact with other subsystems.

[0014] S230: The configuration database stores the configuration state created by the network application, including port configuration, interface, and virtual LAN.

[0015] S240. State Database: This database stores the key operation states of entities configured in the system; this state is used to resolve dependencies between different subsystems; this database stores all the states required to resolve cross-module dependencies.

[0016] S250, network card database storage driver network chip configuration and operation required status;

[0017] S260, the counter database stores counters / statistics related to each port in the system; used to fulfill local requests or provided for remote use;

[0018] S270. If there are multiple network cards on the same device, namespaces can be used to distinguish them.

[0019] S280. The consumer process can obtain data item updates by subscribing to data items.

[0020] The aforementioned computing power routing method for white-box switches, S300, involves a switch state service container running in the switch operating system;

[0021] S310, the Switching State Service container, implements network processing of route updates from upper-layer applications and updates the routing table; at the same time, it also handles port-related configuration and management tasks.

[0022] The S320 exchange state service container listens for and processes network events, and dynamically updates the network state based on these events to maintain the stable and efficient operation of the network.

[0023] The S330 exchange state service container provides an interface for interacting with the database.

[0024] The aforementioned computing power routing method for white-box switches, in S400, includes:

[0025] The S400 and BGP (Border Gateway Protocol) modules, as a type of network application, run independently in a container. The computational routing is calculated in the Border Gateway Protocol container, and the computational routing is based on an extension of the Border Gateway Protocol.

[0026] S410, the calculation of computing power routing requires computing power information, which includes the computing power, memory capacity and storage capacity of computing power nodes.

[0027] The beneficial effects of this invention are:

[0028] The white-box switch computing power routing proposed in this invention has the following advantages, effectively overcoming many of the shortcomings mentioned above:

[0029] (1) It fills the gap in the current white-box switches that lack computing power routing function;

[0030] (2) By introducing computing power routing, the computing resources of each node in the network can be utilized to the maximum extent, thereby improving resource utilization. This optimization capability enables the network to utilize limited resources more efficiently while ensuring performance, thus reducing network operating costs;

[0031] (3) Computing power routing provides important support for distributed computing and edge computing. It can allocate tasks to the most suitable nodes for execution based on the computational load of the task and the computing power requirements of the nodes, thereby realizing the efficient execution and collaboration of distributed computing tasks. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of an architecture for the white-box switch operating system described in this invention.

[0033] Figure 2 This is a schematic diagram of the computing power routing implementation process described in this invention. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] This invention proposes a computing power routing solution for white-box switch network operating systems. This solution captures computing power routing data from the network chip by real-time monitoring of the network interface card (NIC) status and effectively writes this crucial information into the switch's database. Subsequently, the database pushes this computing power routing information to service processes. Upon receiving the computing power routing information from the database, the service processes recalculate the optimal routing path based on this data and update the routing table accordingly, thereby achieving computing power routing.

[0036] Combination Figure 1 and Figure 2 The present invention will be described in detail below.

[0037] The solution monitors the network interface card (NIC) status and effectively records the computing power routing information captured by the network chip into the switch database. Subsequently, the database publishes this critical information to the service processes, enabling them to recalculate routes and update routing table entries based on the computing power routing information obtained from the core database.

[0038] like Figure 2 As shown, a specific scheme for a computing power routing method for white-box switches is as follows:

[0039] The S100 employs a microservices architecture, allowing each functional module of the switch to run in an independent container. This design significantly reduces dependencies and coupling between modules, greatly improving system flexibility and facilitating subsequent functional upgrades, secondary development, and maintenance. In its implementation, the switch's network operating system uses Linux as its underlying operating system platform. Figure 1 As shown.

[0040] The S200 switch's database runs in an independent container, and application layer business modules do not interact directly with the operating system kernel. Instead, they interact with the database, which uniformly publishes information to the listening program, which then interacts with the kernel and hardware. This design ensures the stability of the operating system kernel and reduces direct coupling between business modules.

[0041] The S210 switch's database architecture is divided into five dedicated sub-databases to support the implementation of computing power routing schemes and ensure efficient network operations. These sub-databases include an application layer database, a configuration database, a status database, a network interface card (NIC) database, and a counter database, each with its own unique role and function.

[0042] S220. This database aggregates state information generated by all application containers, such as routing information, next-hop information, and neighbor information, becoming the access point for all applications that wish to interact with other parts of the network. It also stores computing power information entries required for computing power routing, including the computing power, memory capacity, and storage capacity of different computing power nodes. This allows the computing power routing process to calculate a reasonable next-hop route based on this computing power information.

[0043] S230: Configuration states generated by network applications are stored here, including port configurations, interface settings, and VLAN information. This provides a centralized management location for network configuration.

[0044] S240. The state database retains the key operational states of configuration entities in the system. This state information helps to solve the dependency problem between different subsystems and avoids timing problems in the configuration distribution process.

[0045] The S250 network interface card (NIC) database is responsible for recording the status information required for the configuration and operation of network interface cards (NICs), application-specific integrated circuits (ASICs), or network processing units (NPUs), ensuring the correct configuration and efficient operation of the relevant hardware.

[0046] The S260 counter database collects counters and statistics related to each port in the system. This data can meet local monitoring and analysis needs, and can also be accessed remotely to facilitate network administrators in monitoring device status.

[0047] S270. In multi-NIC device scenarios, namespace technology is used to distinguish and manage different NICs independently.

[0048] S270. Consumer processes that need to obtain relevant data from the database can obtain data updates in real time by subscribing to specific data items. This ensures the timeliness and accuracy of information, thereby supporting the rapid response and execution of computing power routing decisions.

[0049] In the implementation of computing power routing in the S300 switch operating system, the switch state service container is the central hub for network processing tasks, playing a crucial role, especially in the execution of computing power routing policies. The switch state service container is responsible for handling network operations, including updating routing decisions, managing port states, and responding to network events in real time.

[0050] S310, the Switching State Service container, handles routing update requests from upper-layer applications (including computing power routing applications), updating the routing table in a timely manner to reflect the latest network topology and path optimization decisions. This container is responsible for forwarding table, aggregation port, neighbor, and port synchronization functions. Furthermore, this container is also responsible for handling port configuration and management tasks.

[0051] The core computation of S400's computing power routing relies on detailed computing power information, including but not limited to the computing power, memory capacity, and storage space of the computing nodes. The Border Gateway Protocol (BGP) related processes are deployed in a separate container, responsible for executing this complex computation process and maintaining and executing the entire routing stack.

[0052] S410 runs the BGP process and related services in a separate container. This not only ensures process isolation and enhances system security and stability, but also improves service maintainability and scalability.

[0053] The main functional modules in this container include:

[0054] 1) Standard Border Gateway Service Process: This process is responsible for handling external routing information, receiving routing status from the outside world through standard TCP (Transmission Control Protocol) or UDP (User Datagram Protocol) sockets. Upon receiving the information, this process effectively pushes the routing information to the forwarding plane through the service interface of the routing manager or forwarding plane manager, ensuring timely updates and accurate transmission of routing information.

[0055] 2) Traditional IP Router Manager Service Process: This service process provides a series of key services, including updating the kernel routing table, interface lookup, and redistribution of routing information across different routing protocols. Through communication with the kernel, it pushes newly calculated forwarding tables to the kernel, and also pushes information to relevant southbound components via the Forwarding Plane Manager (FPM), achieving comprehensive synchronization and consistency management of routing information.

[0056] 3) Forwarding Plane Management Synchronization Service Process: This service is responsible for collecting the routing table status generated by the routing manager and then synchronizing this information to the application layer database within the database engine. This step is crucial for ensuring that computing power routing information is accurately reflected throughout the network system.

[0057] 4) Computing Power Routing Service Process: This process is specifically responsible for handling the core logic of computing power routing. It not only collects computing power information from the network interface card (NIC) but also calculates the routing table based on this information. After calculation, the computing power routing service process synchronizes the updated routing table to the application layer database, thereby guiding the optimized transmission path of data in the network. Example

[0058] Example 1 is a computing power routing function implemented based on the computing power routing method for white-box switches described above.

[0059] The implementation example is deployed in a switching center network, which includes multiple computing nodes, each with varying computing power, memory capacity, and storage space. These nodes are connected via high-speed network switches. The network switches run a network operating system that supports computing power routing, and deploy key components such as a switching state service container, a Border Gateway Protocol (BGP) container, and a database container. Each computing node periodically reports its computing power information (including CPU utilization, available memory, and storage space) to the network switch via its built-in network interface card (NIC). This information is captured and recorded by the computing power routing service process.

[0060] After initial processing, the computing power information is synchronized to the application layer database to facilitate subsequent routing calculations.

[0061] The forwarding plane management synchronization service process synchronizes the current routing table state generated by the route manager to the application layer database.

[0062] The computing power routing service process calculates the optimal data transmission path based on the collected computing power information and the current network status. The calculation process takes into account the load of each computing node, prioritizing nodes with high computing power and low load as data transmission relays.

[0063] Based on the computing power routing calculation results, the routing table entries are updated, and the updated routing information is pushed to the network kernel and forwarding plane through the standard border gateway service process and the IP route manager service process to achieve real-time routing updates.

[0064] During data transmission, network switches guide the forwarding of data packets according to the latest computing power routing table, ensuring that data packets are transmitted through the most powerful and least congested path.

[0065] High-performance computing tasks are prioritized for allocation to nodes with sufficient computing resources, improving task execution efficiency and the overall operational performance of the data center.

[0066] Compared to the routing mode of traditional white-box switches, computing power routing incorporates computing power as a routing calculation dimension when selecting forwarding paths, giving it a greater advantage in performance-sensitive application scenarios, especially in modern network environments that require large amounts of data processing and computation.

[0067] The introduction of computing power routing provides a better network infrastructure for cloud computing, artificial intelligence, vehicle-to-everything (V2X) and Internet of Things (IoT) applications.

[0068] The embodiments described above can be further combined or replaced, and these embodiments are merely descriptions of preferred embodiments of the present invention, not limitations on the concept and scope of the present invention. Various changes and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the inventive concept are all within the protection scope of the present invention. The protection scope of the present invention is given by the appended claims and any equivalents.

Claims

1. A computing power routing method for white-box switches, characterized in that, By monitoring the status of the network card and writing the computing power routing information captured by the network chip into the switch database, the database publishes the relevant information to the business process. After obtaining the computing power routing information from the database, the business process recalculates the route and updates the routing table entries, thereby realizing computing power routing. The computing power routing method for white-box switches includes: The S100 and its switches operate in independent containers, employing a microservice architecture to reduce coupling between business modules and facilitate functional upgrades and secondary development. The S200 database, as a core component, stores the status information of the switch and serves as a middleware for message passing, running independently in a container. Information and data transmission between service processes are carried out through the database. The producer process writes data into the database, and then the database notifies the consumer process listening for the event. The consumer process retrieves the necessary information and data from the database. S300, within the network operating system, runs a switching state service container, which undertakes network processing tasks; The calculation of S400 and computing power routing is carried out in the border gateway protocol container, and computing power routing is based on the extended implementation of the border gateway protocol. S200 includes the following: S210. The database runs in a database container. The database is divided into application database, configuration database, status database, network card data, and counter database. S220, the application database stores the state generated by all application containers, including computing power information, routing information, next hop, and neighbors; it is the southbound entry point for all applications that wish to interact with other subsystems. S230: The configuration database stores the configuration state created by the network application, including port configuration, interface, and virtual LAN. S240. State Database: This database stores the key operation states of entities configured in the system; this state is used to resolve dependencies between different subsystems; this database stores all the states required to resolve cross-module dependencies. S250, network card database storage driver network chip configuration and operation required status; S260, the counter database stores counters / statistics related to each port in the system; used to fulfill local requests or provided for remote use; S270. If there are multiple network cards on the same device, namespaces can be used to distinguish them. S280. The consumer process can obtain data item updates by subscribing to data items.

2. The computing power routing method for a white-box switch according to claim 1, characterized in that, S300, within the network operating system, runs a switching state service container, which undertakes network processing tasks; The exchange state service container implements network processing for route updates from upper-layer applications and updates the routing table; it also handles port-related configuration and management tasks. The exchange state service container listens for and processes network events, and dynamically updates the network state based on these events to maintain the stable and efficient operation of the network. The exchange state service container provides an interface for interacting with the database.

3. The computing power routing method for a white-box switch according to claim 1, characterized in that, The S400 includes: The Border Gateway Protocol (BGP) module, as a network application, runs independently in a container. The computational routing is calculated within the BGP container, and the computational routing is based on an extension of the BGP protocol. The calculation of computing power routing requires computing power information, which includes the computing power, memory capacity, and storage capacity of the computing power nodes.

Images