Network traffic forwarding method and device, electronic equipment and storage medium

Abstract

The invention provides a network traffic forwarding method and device, electronic equipment and a storage medium, and is applied to an intelligent network card, the method comprises the following steps: obtaining link quality information measured by the intelligent network card for a cross-cloud link; reporting the link quality information to a forwarding gateway in the cloud; receiving a path selection strategy issued by the forwarding gateway, the path selection strategy being generated by the forwarding gateway based on the link quality information and preset strategy information; and forwarding the network traffic according to the path selection strategy. Therefore, real-time perception of cross-cloud link quality and quick path selection based on real-time quality are realized, the problem of path selection strategy lag caused by the fact that detection and decision-making functions are concentrated on a control level in the background technology is fundamentally solved, and the quality and real-time performance of cross-cloud network transmission are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cloud network, and particularly relates to a network traffic forwarding method and device, electronic equipment and storage medium. BACKGROUND

[0002] With the migration of enterprise business to multi-cloud and hybrid cloud architecture, cross-cloud enterprise network gradually becomes the mainstream way to connect different cloud service providers and private data centers. Under this architecture, network traffic needs to pass through multiple links with different transmission characteristics. These links have significant differences in performance indicators such as latency, bandwidth, and packet loss rate, and their quality will dynamically change with network conditions. Traditional network path selection methods are difficult to adapt to this complex and variable inter-cloud network environment, and cannot provide stable quality of service guarantee for applications sensitive to network performance.

[0003] Currently, there are two main ways of path selection in cross-cloud network: one is static configuration-based policy routing, in which network administrators pre-set the priority and weight of each link, and traffic is forwarded according to fixed policies; the other is Border Gateway Protocol (BGP) -based dynamic routing, which automatically selects the shortest path by exchanging reachability information between autonomous systems.

[0004] However, these existing solutions have a common technical problem: since the link quality detection and path decision functions are concentrated in the control plane, they cannot real-time perceive the dynamic quality changes of cross-cloud links, resulting in path selection strategy lag and difficulty in meeting the demand for high-quality transmission. SUMMARY

[0005] The present application provides a network traffic forwarding method, device, electronic equipment and storage medium to solve the problem that in the prior art, since the link quality detection and path decision functions are concentrated in the control plane, they cannot real-time perceive the dynamic quality changes of cross-cloud links, resulting in path selection strategy lag and difficulty in meeting the demand for high-quality transmission.

[0006] In a first aspect, the present application provides a network traffic forwarding method applied to an intelligent network card, comprising: obtaining link quality information measured by the intelligent network card for a cross-cloud link; reporting the link quality information to a forwarding gateway in the cloud; receiving a path selection strategy issued by the forwarding gateway, wherein the path selection strategy is generated by the forwarding gateway based on the link quality information and pre-set strategy information; forwarding network traffic according to the path selection strategy.

[0007] In a possible implementation, the obtaining the link quality information measured by the intelligent network card for the cross-cloud link comprises: generating a link probe packet according to a preset probe packet format every preset time interval; sending the link probe packet to a peer forwarding gateway through the cross-cloud link; receiving a response packet returned by the peer forwarding gateway; determining link quality information of the cross-cloud link based on a sending time of the link probe packet and a receiving time of the response packet, the link quality information comprising at least one of a transmission delay, a delay jitter value and a packet loss rate.

[0008] In a possible implementation, the method further comprises: monitoring a network data flow through the intelligent network card; generating a burst flow event containing flow identification information in a case where a flow rate of the network data flow is identified to exceed a preset threshold; reporting the burst flow event to the forwarding gateway, wherein the path selection strategy is generated by the forwarding gateway in response to the burst flow event.

[0009] In a possible implementation, the forwarding network traffic according to the path selection strategy comprises: determining a forwarding path of a data flow corresponding to the burst flow event according to the path selection strategy; creating a flow entry in a hardware forwarding table of the intelligent network card according to a mapping relationship between the flow identification information and the forwarding path; matching a packet feature of the network traffic with the flow entry in the hardware forwarding table; if the matching is successful, completing the forwarding of the network traffic in the intelligent network card according to a forwarding path specified in the matched flow entry; if the matching fails, reporting flow identification information corresponding to the network traffic to the forwarding gateway, and receiving a forwarding instruction issued by the forwarding gateway based on the flow identification information, and forwarding the network traffic according to the forwarding instruction.

[0010] In a possible implementation, the method further comprises: receiving a path switching instruction issued by the forwarding gateway, the path switching instruction being used to indicate switching of a forwarding path of a target data flow to a new path, the path switching instruction containing flow identification information of the target data flow and a corresponding new path; updating forwarding path information corresponding to the flow identification information in a local policy table to the new path; For subsequent network traffic that belongs to the target data stream, the network traffic is forwarded according to the new path updated in the local policy table.

[0011] In one possible implementation, after updating the forwarding path information corresponding to the flow identifier information in the local policy table to the new path, the method further includes: Within a preset switching time window, each network traffic packet belonging to the target data stream is copied and forwarded via the original path and the new path respectively. After the switching time window ends, network traffic forwarding the target data stream to the original path is stopped.

[0012] In one possible implementation, the path switching instruction is also used to instruct session state synchronization of the target data stream; The method further includes: In response to the path switching command, communicate with the forwarding gateway to obtain the current session state information of the target data stream; During the forwarding of network traffic according to the new path updated in the local policy table, the network traffic is processed based on the current session state information to ensure that the target data stream session state is normal.

[0013] In one possible implementation, the method further includes: Monitor the hardware acceleration function status of the smart network card; When the hardware acceleration function is detected to be in an abnormal state, a function abnormality notification is sent to the forwarding gateway. Stop executing the step of forwarding network traffic according to the path selection strategy, and upload the received network traffic to the forwarding gateway; In response to the functional error notification, the forwarding gateway switches to software forwarding mode to process the uploaded network traffic.

[0014] In one possible implementation, the link quality information includes multiple quality parameters, and the preset policy information includes tenant service level agreement requirements. The path selection strategy is generated by the forwarding gateway by performing the following operations: The weight parameters corresponding to each of the quality parameters are determined according to the requirements of the tenant service level agreement. The quality parameters are weighted according to the weight parameters corresponding to each quality parameter to obtain the comprehensive score of each optional path; The path with the highest overall score is selected from the available paths as the designated forwarding path, and the path selection strategy is generated based on the designated forwarding path.

[0015] In one possible implementation, forwarding network traffic according to the path selection strategy includes: Identify the service category corresponding to the network traffic; The network traffic is forwarded according to the path selection strategy and the service category.

[0016] Secondly, this application provides a network traffic forwarding device applied to a smart network interface card (NIC), comprising: The acquisition module is used to acquire the link quality information measured by the smart network card for cross-cloud links; The reporting module is used to report the link quality information to the forwarding gateway in the cloud; The receiving module is used to receive the path selection policy issued by the forwarding gateway, wherein the path selection policy is generated by the forwarding gateway based on the link quality information and preset policy information; The forwarding module is used to forward network traffic according to the path selection strategy.

[0017] In one possible implementation, the acquisition module is specifically used for: At preset time intervals, link probe messages are generated according to a preset probe message format; The link probe message is sent to the peer forwarding gateway through the cross-cloud link; Receive the response message returned by the peer forwarding gateway; Based on the sending time of multiple link probe messages and the receiving time of the response messages, the link quality information of the cross-cloud link is determined. The link quality information includes at least one of transmission delay, delay jitter value, and packet loss rate.

[0018] In one possible implementation, the device further includes a monitoring module for: Monitor network data flow through the smart network card; If the network data flow rate is found to exceed a preset threshold, a burst flow event containing flow identification information is generated; The burst flow event is reported to the forwarding gateway, wherein the path selection strategy is generated by the forwarding gateway in response to the burst flow event.

[0019] In one possible implementation, the forwarding module is specifically used for: Based on the path selection strategy, determine the forwarding path of the data stream corresponding to the burst event; Based on the mapping relationship between the flow identifier information and the forwarding path, a flow table entry is created in the hardware forwarding table of the smart network card; For the received network traffic, the packet characteristics of the network traffic are matched with the flow table entries in the hardware forwarding table; If a match is successful, the network traffic is forwarded on the smart network interface card according to the forwarding path specified in the matched flow table entry. If the matching fails, the flow identification information corresponding to the network traffic is reported to the forwarding gateway, and the forwarding gateway receives the forwarding instruction issued by the forwarding gateway based on the flow identification information, and forwards the network traffic according to the forwarding instruction.

[0020] In one possible implementation, the device further includes a switching module for: The system receives a path switching instruction from the forwarding gateway. The path switching instruction is used to indicate that the forwarding path of the target data stream be switched to a new path. The path switching instruction includes the flow identification information of the target data stream and the corresponding new path. Update the forwarding path information in the local policy table corresponding to the flow identifier information to the new path; For subsequent network traffic that belongs to the target data stream, the network traffic is forwarded according to the new path updated in the local policy table.

[0021] In one possible implementation, the apparatus further includes a replication module for: Within a preset switching time window, each network traffic packet belonging to the target data stream is copied and forwarded via the original path and the new path respectively. After the switching time window ends, network traffic forwarding the target data stream to the original path is stopped.

[0022] In one possible implementation, the path switching instruction is further configured to instruct session state synchronization of the target data stream; the apparatus further includes a processing module configured to: In response to the path switching command, communicate with the forwarding gateway to obtain the current session state information of the target data stream; During the forwarding of network traffic according to the new path updated in the local policy table, the network traffic is processed based on the current session state information to ensure that the target data stream session state is normal.

[0023] In one possible implementation, the apparatus further includes a transmitting module for: Monitor the hardware acceleration function status of the smart network card; When the hardware acceleration function is detected to be in an abnormal state, a function abnormality notification is sent to the forwarding gateway. Stop executing the step of forwarding network traffic according to the path selection strategy, and upload the received network traffic to the forwarding gateway; In response to the functional error notification, the forwarding gateway switches to software forwarding mode to process the uploaded network traffic.

[0024] In one possible implementation, the link quality information includes multiple quality parameters, and the preset policy information includes tenant service level agreement requirements. The path selection strategy is generated by the forwarding gateway by performing the following operations: The weight parameters corresponding to each of the quality parameters are determined according to the requirements of the tenant service level agreement. The quality parameters are weighted according to the weight parameters corresponding to each quality parameter to obtain the comprehensive score of each optional path; The path with the highest overall score is selected from the available paths as the designated forwarding path, and the path selection strategy is generated based on the designated forwarding path.

[0025] In one possible implementation, the forwarding module is specifically used for: Identify the service category corresponding to the network traffic; The network traffic is forwarded according to the path selection strategy and the service category.

[0026] Thirdly, this application provides an apparatus comprising: a processor and a memory, the processor being configured to execute a network traffic forwarding program stored in the memory to implement the network traffic forwarding method described in any one of the first aspects.

[0027] Fourthly, this application provides a storage medium storing one or more programs that can be executed by one or more processors to implement the network traffic forwarding method described in any one aspect.

[0028] Compared with the prior art, the technical solution provided in this application has the following advantages: First, the method provided in this application obtains real-time link quality information for cross-cloud links through the smart network interface card (NIC). This step decentralizes the quality detection function from the control layer to the data layer NIC, solving the latency problem caused by centralized detection. Second, the measured real-time quality information is reported to the forwarding gateway, providing an accurate and timely data foundation for path decision-making. Then, the path selection strategy generated by the forwarding gateway based on this real-time quality information and preset strategies is received. This step ensures that the decision is based on both real-time network conditions and business requirements. Finally, traffic is forwarded directly according to the strategy, achieving rapid response of the data plane based on real-time strategies. This achieves real-time perception of cross-cloud link quality and rapid path selection based on real-time quality, fundamentally solving the problem of delayed path selection strategies caused by the concentration of detection and decision-making functions at the control layer in the background technology, and significantly improving the quality and real-time performance of cross-cloud network transmission. Attached Figure Description

[0029] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0030] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0031] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0032] Figure 1 A flowchart illustrating an embodiment of a network traffic forwarding method provided in this application; Figure 2 A flowchart illustrating an embodiment of another network traffic forwarding method provided in this application; Figure 3 A flowchart illustrating another embodiment of a network traffic forwarding method provided in this application; Figure 4 A network traffic forwarding system architecture diagram provided in this application embodiment; Figure 5 A block diagram illustrating an embodiment of a network traffic forwarding device provided in this application; Figure 6This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0034] The following disclosure provides numerous different embodiments or examples for implementing various structures of this application. To simplify the disclosure, specific examples of components and arrangements are described below. These are merely examples and are not intended to limit the scope of this application. Furthermore, reference numerals and / or letters may be repeated in different examples. Such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.

[0035] To address the technical problem in existing technologies where link quality detection and path decision-making functions are concentrated at the control layer, making it impossible to perceive dynamic quality changes of cross-cloud links in real time, resulting in lagging path selection strategies and difficulty in meeting the requirements of high-quality transmission, this application provides a network traffic forwarding method that enables real-time perception of cross-cloud link quality and rapid path selection based on real-time quality. This fundamentally solves the problem of lagging path selection strategies caused by the concentration of detection and decision-making functions at the control layer in the background technology, and significantly improves the quality and real-time performance of cross-cloud network transmission.

[0036] Figure 1 This is a flowchart illustrating an embodiment of a network traffic forwarding method provided in this application. In one embodiment, the method is applied to a smart network interface card (NIC). Figure 1 As shown, the method includes the following steps: Step 101: Obtain the link quality information measured by the smart network card for cross-cloud links.

[0037] Smart network interface card (NIC): refers to a network adapter with programmable capabilities and a hardware acceleration engine. In addition to performing standard network transmission functions, it also provides a built-in programmable and configurable hardware acceleration engine.

[0038] Cross-cloud link: refers to the network transmission channel that connects different cloud platforms or data centers.

[0039] Link quality information refers to metrics that reflect the transmission performance of a network link, including performance indicators such as transmission delay, delay jitter, and packet loss rate.

[0040] In this embodiment of the application, step 101 may specifically include the following steps: generating a link probe message according to a preset probe message format at preset time intervals; sending the link probe message to the peer forwarding gateway through the cross-cloud link; receiving a response message returned by the peer forwarding gateway; determining the link quality information of the cross-cloud link based on the sending time of multiple link probe messages and the receiving time of the response message, wherein the link quality information includes at least one of transmission delay, delay jitter value and packet loss rate.

[0041] The preset time interval refers to the pre-configured detection period time parameter, ranging from 100ms to 10s, which can be adjusted according to network stability requirements; the detection message format refers to the data packet structure specifically defined for link quality detection, including message type, sequence number, and timestamp fields; the peer forwarding gateway refers to the forwarding gateway device located at the other end of the cross-cloud link and establishing a communication session with the local gateway; the transmission delay refers to the one-way or round-trip time delay experienced by a data packet from sending to receiving; the delay jitter value refers to the standard deviation of the change in transmission delay over multiple consecutive transmissions; and the packet loss rate refers to the ratio of the number of lost data packets to the total number of sent packets within the statistical period.

[0042] In this scheme, the hardware timer in the smart network interface card (NIC) triggers the probe message generation module at preset time intervals to construct a probe message containing an incrementing sequence number and a nanosecond-level sending timestamp. The message is sent to the peer forwarding gateway via a dedicated queue through a cross-cloud link. Upon receiving the message, the peer gateway immediately generates a response message containing the receiving timestamp and the original sending timestamp. The local smart NIC records the arrival time of the response message and extracts the timestamp information. The transmission delay is calculated using the formula: Delay = (Receiving timestamp - Sending timestamp) / 2, taking the arithmetic mean of the most recent N measurements. The delay jitter is calculated using the formula: Jitter = √[Σ(Delay i - Average Delay)² / (N-1)], which is the sample standard deviation. The packet loss rate is calculated using the formula: Packet loss rate = (Number of packets sent - Number of packets received) / Number of packets sent × 100%, with the statistical window being the most recent M probe cycles.

[0043] This solution utilizes the hardware timestamps and dedicated processing engine of the smart network interface card to achieve microsecond-level precision in measuring link quality parameters. Through rigorous mathematical calculations and statistical processing, it ensures the accuracy and reliability of the quality parameters, providing precise data input for subsequent intelligent path selection. This effectively overcomes the shortcomings of insufficient accuracy and excessive overhead in traditional software detection, and meets the technical requirements for link status awareness in high-quality cross-cloud network transmission.

[0044] Step 102: Report the link quality information to the forwarding gateway in the cloud.

[0045] Reporting: refers to the process of sending locally acquired data to a designated receiving end via network transmission.

[0046] Forwarding gateway in the cloud: refers to a gateway device deployed in a cloud computing environment that undertakes the functions of network traffic forwarding and routing decisions.

[0047] In this embodiment, the smart network interface card (NIC) encapsulates the link quality information (including parameters such as transmission latency, latency jitter, and packet loss rate) measured in step 101 according to a predefined JSON format. The encapsulated data includes fields such as timestamp, link identifier, and quality parameter values. After encapsulation, the quality information data is sent to a designated listening port of the forwarding gateway via a dedicated management network interface using UDP (User Datagram Protocol). The sending process employs a periodic reporting mechanism, sending data at fixed time intervals, and immediately triggering an emergency report when a sudden change in link quality is detected. After receiving the data, the forwarding gateway parses the JSON-formatted quality information, updates the link state database, and provides real-time data support for the generation of path selection strategies.

[0048] In this way, an efficient quality information transmission mechanism was established between the smart network interface card and the forwarding gateway, ensuring the timeliness and accuracy of link quality data, providing a reliable data foundation for subsequent intelligent path decision-making, and reducing the impact on the service channel through a dedicated management network and optimized data transmission methods.

[0049] Step 103: Receive the path selection policy issued by the forwarding gateway, wherein the path selection policy is generated by the forwarding gateway based on the link quality information and preset policy information.

[0050] Path selection strategy: refers to a set of instructions that includes rules for selecting the path for forwarding data streams.

[0051] Pre-configured policy information: refers to pre-configured business policy requirements, including SLA (Service Level Agreement) requirements, etc.

[0052] In this embodiment, the forwarding gateway generates an optimal path selection strategy based on the received link quality information and preset policy requirements using algorithms such as weighted calculation, and then sends this strategy to the smart network interface card (NIC). The smart NIC receives and parses these policy instructions and stores them in its local policy table.

[0053] Specifically, the link quality information includes multiple quality parameters, the preset policy information includes tenant service level agreement requirements, and the path selection policy is generated by the forwarding gateway by performing the following operations: determining the weight parameters corresponding to each quality parameter according to the tenant service level agreement requirements; performing a weighted calculation on all quality parameters according to the weight parameters corresponding to each quality parameter to obtain a comprehensive score for each optional path; selecting the path with the highest comprehensive score from all optional paths as the designated forwarding path, and generating the path selection policy based on the designated forwarding path.

[0054] Weighting parameters refer to coefficient values ​​used to reflect the importance of different quality parameters in the scoring calculation, with values ​​ranging from 0 to 1; weighted calculation refers to the calculation method of multiplying each quality parameter by its corresponding weight and then summing the results; comprehensive score refers to the quantitative score representing the overall quality of the path obtained through weighted calculation; designated forwarding path refers to the optimal data transmission path determined based on the scoring results.

[0055] In this scheme, the forwarding gateway first parses the tenant's SLA requirements and determines the weight allocation of each quality parameter according to the service type (e.g., assigning higher weights to latency parameters for latency-sensitive services). Then, it calculates the weighted sum of quality parameters for each optional path: Overall Score = Σ(Parameter Value × Weight Coefficient), where latency and packet loss rate are used in reciprocal form to ensure numerical consistency. Finally, it selects the path with the highest overall score as the optimal path and generates a path selection strategy containing path identifiers and priority information. For example, the latency weight α is 0.6, the jitter weight β is 0.3, and the packet loss rate weight γ is 0.1. The calculation formula is: Score = α × (1 / Latency) + β × (1 / Jitter) + γ × (1 / Packet Loss Rate).

[0056] This solution enables quantitative path evaluation based on multi-dimensional quality parameters and tenant SLA requirements. Through scientific weighted calculation and comprehensive scoring mechanism, it ensures that path selection takes into account both real-time network conditions and business quality requirements, effectively improving the quality assurance capability and resource utilization of cross-cloud network transmission.

[0057] Step 104: Forward network traffic according to the path selection strategy.

[0058] In this embodiment of the application, step 104 may specifically include the following steps: identifying the service category corresponding to the network traffic; and forwarding the network traffic according to the path selection strategy and the service category.

[0059] Service category refers to the category divided according to the characteristics of network traffic and application type, including different types such as real-time audio and video, database synchronization, and file transfer; path selection strategy refers to the set of instructions containing forwarding rules and priority settings, which is generated by the forwarding gateway based on link quality and SLA requirements.

[0060] In this embodiment, the smart network interface card (NIC) identifies the service category of network traffic through DPI (deep packet inspection) technology or traffic feature analysis based on the five-tuple (source IP, destination IP, source port, destination port, and protocol type). Then, according to the forwarding rules for different service categories in the received path selection strategy (e.g., real-time audio and video traffic prioritizes low-latency paths, and large data transmission traffic prioritizes high-bandwidth paths), it selects the corresponding forwarding path for each data stream. Finally, the smart NIC's hardware forwarding engine performs packet encapsulation and forwarding operations according to the determined path.

[0061] This solution enables refined traffic scheduling based on business characteristics, allowing the selection of the optimal forwarding path according to the performance requirements of different services. This not only ensures the service quality of critical services but also improves the utilization of network resources, effectively enhancing the overall performance and user experience of cross-cloud network transmission.

[0062] The technical solution provided in this application firstly acquires real-time link quality information for cross-cloud links through the smart network interface card (NIC). This step decentralizes the quality detection function from the control layer to the data layer of the smart NIC, solving the latency problem caused by centralized detection. Secondly, the measured real-time quality information is reported to the forwarding gateway, providing an accurate and timely data foundation for path decision-making. Then, the path selection policy generated by the forwarding gateway based on this real-time quality information and pre-defined policies is received. This step ensures that the decision is based on both real-time network conditions and business requirements. Finally, traffic is forwarded directly according to the policy, achieving rapid response of the data plane based on real-time policies. This achieves real-time perception of cross-cloud link quality and rapid path selection based on real-time quality, fundamentally solving the problem of delayed path selection policies caused by the concentration of detection and decision-making functions at the control layer in the background technology, and significantly improving the quality and real-time performance of cross-cloud network transmission.

[0063] Figure 2 A flowchart illustrating another embodiment of a network traffic forwarding method provided in this application. Figure 2 The process shown is in Figure 1 Based on the illustrated process, the following steps are included: Step 201: Monitor the network data flow through the smart network card.

[0064] Step 202: If the network data flow rate is found to exceed a preset threshold, a burst flow event containing flow identification information is generated.

[0065] Step 203: Report the burst flow event to the forwarding gateway, wherein the path selection strategy is generated by the forwarding gateway in response to the burst flow event.

[0066] For ease of understanding, steps 201-203 will be explained uniformly below: Network data flow: refers to the set of packets that can be uniquely identified and processed at the hardware level of a smart network interface card (NIC). It is typically segmented based on the five-tuple (source IP address, destination IP address, source port number, destination port number, and transport layer protocol type) characteristics in the packet header. The smart NIC uses a built-in flow classifier (such as TCAM or a hardware lookup engine) to match and distinguish different flows in real time, maintaining an independent hardware counter and status information for each flow. Traffic rate: refers to the rate at which data traffic passes through the smart NIC per unit of time, usually measured in bits per second (bps) or packets per second (pps). Preset threshold: refers to a pre-configured rate threshold value used to identify burst flows; different thresholds can be set according to different service types. Burst flow event: refers to an event notification generated when an abnormal traffic rate is detected, containing flow identifier and rate information. Flow identifier information: refers to the characteristic information that uniquely identifies a data flow, including five-tuple information and flow statistical characteristics.

[0067] In this embodiment, the traffic monitoring module in the smart network interface card (NIC) continuously monitors the traffic rate of each data stream passing through, and calculates the average rate over the most recent time period using a sliding time window algorithm (e.g., calculated every 5 seconds). When the rate of a data stream exceeds a preset threshold (e.g., 100Mbps or 5000pps), it is determined to be a burst event. Subsequently, a burst event report containing the data stream's five-tuple identifier, current rate value, timestamp, and other information is generated. Finally, the event report is reported to the forwarding gateway through a dedicated control channel. Upon receiving the burst event, the forwarding gateway generates a dedicated path selection strategy for the burst event based on the current network conditions and policy requirements.

[0068] pass Figure 2 The process shown enables real-time detection and rapid response to network traffic bursts, timely identification of abnormal traffic and triggering path optimization strategies, effectively avoiding network congestion and performance degradation, ensuring the service quality of critical services, and improving network resource utilization efficiency and overall transmission performance.

[0069] exist Figure 2Based on the illustrated process, step 104 may specifically include the following steps: determining the forwarding path of the data flow corresponding to the burst flow event according to the path selection strategy; creating a flow table entry in the hardware forwarding table of the smart network card according to the mapping relationship between the flow identifier information and the forwarding path; for the received network traffic, matching the packet characteristics of the network traffic with the flow table entry in the hardware forwarding table; if the match is successful, forwarding the network traffic on the smart network card according to the forwarding path specified in the matched flow table entry; if the match fails, reporting the flow identifier information corresponding to the network traffic to the forwarding gateway, and receiving the forwarding instruction issued by the forwarding gateway based on the flow identifier information, and forwarding the network traffic according to the forwarding instruction.

[0070] Hardware forwarding table refers to the dedicated hardware storage structure integrated inside the smart network interface card (NIC), such as TCAM (Ternary Content-Addressable Memory) or SRAM (Static Random-Access Memory), used for high-speed storage of flow table entries; flow table entry refers to the forwarding rule entry stored in the hardware forwarding table, containing flow identification information (such as a 5-tuple) and the corresponding forwarding action; packet characteristics refer to the basic information in the packet header used for flow identification, typically including source IP address, destination IP address, source port number, destination port number, and protocol type; forwarding command refers to the control command generated by the forwarding gateway that contains specific forwarding path information.

[0071] In this embodiment, when the smart network interface card (NIC) identifies and reports a burst flow event, it receives the path selection policy for the burst flow from the forwarding gateway; determines the optimal forwarding path for the data flow according to the policy; then creates a new flow table entry in the hardware forwarding table of the smart NIC, establishing a mapping relationship between flow identification information and forwarding path; for subsequent received network traffic, the smart NIC uses a hardware lookup engine to perform parallel matching of packet characteristics with flow table entries: if the match is successful, the packet modification and forwarding are completed directly at the hardware level (such as adding a VXLAN (Virtual Extensible Local Area Network) header or modifying the next-hop address); if the match fails (i.e., the first packet or a non-burst flow), the flow identification information is reported to the forwarding gateway, and the control plane generates forwarding instructions before execution of forwarding.

[0072] This solution significantly improves the data plane's processing performance and forwarding efficiency by leveraging the hardware lookup and processing capabilities of the smart network interface card. At the same time, through the collaboration between the control plane and the data plane, it ensures the accuracy and flexibility of forwarding decisions, effectively reducing forwarding latency and CPU (Central Processing Unit) overhead.

[0073] Figure 3 A flowchart illustrating another embodiment of a network traffic forwarding method provided in this application. Figure 3 The process shown is in Figure 1 Based on the illustrated process, the following steps are included: Step 301: Receive a path switching instruction issued by the forwarding gateway. The path switching instruction is used to indicate that the forwarding path of the target data stream be switched to a new path. The path switching instruction includes the flow identification information of the target data stream and the corresponding new path.

[0074] Step 302: Update the forwarding path information corresponding to the flow identifier information in the local policy table to the new path.

[0075] Step 303: For subsequent received network traffic belonging to the target data stream, forward the network traffic according to the updated new path in the local policy table.

[0076] For ease of understanding, steps 301-303 will be explained uniformly below: Path switching instruction refers to the control signaling generated by the forwarding gateway to instruct a specific data flow to change its forwarding path; target data flow refers to the specific network data flow that needs to perform path switching operation, which is uniquely identified by the flow identification information; new path refers to the forwarding path identifier specified after the switch; local policy table refers to the policy database stored locally on the smart network interface card that maintains the mapping relationship between flow identifiers and forwarding paths.

[0077] In this embodiment, after receiving a path switching instruction from the forwarding gateway (the instruction includes the five-tuple identifier of the target data flow and the identifier of the new path to be switched), the smart network interface card (NIC) parses the path switching instruction. Then, the NIC queries the existing entries corresponding to the target data flow in its locally maintained policy table and updates the forwarding path field to the new path identifier. For subsequent network traffic, the NIC identifies packets belonging to the target data flow through a hardware flow classifier, queries the updated policy table, and performs forwarding operations according to the new path identifier, including corresponding packet encapsulation and next-hop address modification.

[0078] Figure 3The process shown ensures low latency and high efficiency in path switching through the local policy maintenance and hardware acceleration capabilities of the smart network interface card, effectively avoiding service quality degradation caused by changes in network conditions, and improving the reliability and adaptability of cross-cloud network transmission.

[0079] In another embodiment of this application, after updating the forwarding path information corresponding to the flow identifier information in the local policy table to the new path, the following steps may be included: within a preset switching time window, each network traffic packet belonging to the target data flow is copied and forwarded through the original path and the new path respectively; after the switching time window ends, the forwarding of the network traffic of the target data flow to the original path is stopped.

[0080] The switching time window refers to the path overlap transmission period set to ensure service continuity, usually set to 2-3 times the RTT (Round-Trip Time); packet duplication refers to the processing mechanism of generating an identical copy of the original data packet and forwarding it through different paths simultaneously; the original path refers to the transmission path that the target data stream was using before the switch.

[0081] In this embodiment, after receiving the path switching instruction and updating the local policy table, the smart network interface card starts a timer to calculate the switching time window (typically 100-500 milliseconds). Within this time window, a copy operation is performed on each arriving packet belonging to the target data stream to generate two identical packet copies, which are forwarded in parallel through the original path and the newly designated path, respectively. At the same time, the transmission quality of the new path is monitored. When the time window ends, the sending of packets to the original path is immediately stopped, and the system is completely switched to the new path for unidirectional transmission.

[0082] This solution effectively avoids transmission problems such as out-of-order packets and packet loss that may occur due to path switching by using packet replication and dual-path parallel transmission mechanisms, ensuring the continuity and stability of critical business sessions and significantly improving the reliability of cross-cloud network services and user experience.

[0083] In another embodiment of this application, the path switching instruction is further used to instruct session state synchronization of the target data stream, and the method may further include the following steps: in response to the path switching instruction, communicating with the forwarding gateway to obtain the current session state information of the target data stream; and processing the network traffic based on the current session state information during the forwarding of the network traffic according to the new path updated in the local policy table, so as to ensure that the session state of the target data stream is normal.

[0084] Session state synchronization refers to the mechanism that maintains the consistency of the transport layer session state during path switching, ensuring that the connection is not interrupted due to path changes; current session state information refers to connection state parameters including TCP (Transmission Control Protocol) sequence number, acknowledgment number, window size, etc.; normal session state means that the data stream maintains its original connection state and transmission characteristics after path switching.

[0085] In this embodiment, when the path switching instruction received by the smart network interface card (NIC) includes a session synchronization indication, it immediately sends a session status query request to the forwarding gateway. The forwarding gateway extracts the current status information of the target data stream (including TCP sequence number, timestamp, window parameters, etc.) from its session status table and sends it to the smart NIC. When the smart NIC subsequently forwards the data stream packets through the new path, it adaptively adjusts the packet header based on the acquired status information (such as updating the sequence number, recalculating the checksum, etc.) to ensure that the receiving end can seamlessly continue the data stream. This solution, through the collaborative cooperation of the smart NIC and the forwarding gateway, ensures the continuity and consistency of the packet sequence before and after the switch, effectively avoiding TCP retransmission or connection reset problems caused by path changes, guaranteeing the continuous transmission and service quality of critical services, and improving the reliability and user experience of cross-cloud network transmission.

[0086] In another embodiment of this application, the method may further include the following steps: monitoring the hardware acceleration function status of the smart network card; when the hardware acceleration function status is detected to be abnormal, sending a function abnormality notification to the forwarding gateway; stopping the execution of the step of forwarding network traffic according to the path selection strategy, and uploading the received network traffic to the forwarding gateway; wherein, in response to the function abnormality notification, the forwarding gateway switches to software forwarding mode to process the uploaded network traffic.

[0087] Hardware acceleration function status refers to the working status of the dedicated processing engine (such as programmable pipeline, hardware lookup unit, etc.) in the smart network card, including normal, abnormal, and fault states; Functional abnormality notification refers to the status alarm message sent by the smart network card to the forwarding gateway when it detects a hardware abnormality; Software forwarding mode refers to the backup working mode in which the CPU of the forwarding gateway performs packet forwarding in software when the smart network card is unavailable.

[0088] In this embodiment, the health monitoring module built into the smart NIC periodically checks the status register and working counter of the hardware acceleration engine. When a hardware error flag or abnormal performance indicator is detected, a functional anomaly notification containing detailed fault information is immediately generated and sent to the forwarding gateway via the out-of-band management channel. Subsequently, the smart NIC stops all hardware acceleration processing and directly uploads the received network traffic to the software processing unit of the forwarding gateway through the PCIe (Peripheral Component Interconnect Express) interface. After receiving the anomaly notification, the forwarding gateway starts the software forwarding mode, and its software forwarding components such as DPDK (Data Plane Development Kit) take over the traffic processing task. This solution, through real-time status monitoring and rapid mode switching, ensures that the system can seamlessly fall back to the software forwarding mode when the smart NIC malfunctions, effectively avoiding service interruptions caused by single points of failure, greatly improving the reliability and availability of the system, and ensuring the business continuity of cross-cloud network services.

[0089] Figure 4 This is a network traffic forwarding system architecture diagram provided for an embodiment of this application. Figure 4 As shown, the architecture includes the following components: 1.1 The Cloud Enterprise Network Control Center, as the global control plane, is responsible for unified policy management and scheduling optimization. Specifically, it includes: implementing policy-driven intelligent scheduling based on SLAs, dynamically calculating the optimal path based on real-time link quality indicators (bandwidth / latency / packet loss rate); supporting the issuance of multi-dimensional policies (security policies, traffic scheduling policies, service quality policies) to achieve unified management and control across VPCs (Virtual Private Clouds) and IDCs (Internet Data Centers); and working in conjunction with intelligent detectors to obtain network topology and path health status in real time, forming a closed-loop control mechanism of detection-decision-optimization.

[0090] 1.2 The cloud VPC network domain, as the environment for carrying user services, has the following characteristics: it supports multiple deployment environments such as public cloud, private cloud and IDC, and achieves cross-cloud interconnection through a unified cloud enterprise network architecture; it provides multi-tenant isolation and policy-based interoperability capabilities to ensure security and compliance between different VPCs.

[0091] 1.3 Terminal Gateway (TGW), as a core node of the data plane, is built on DPDK and provides: high-performance routing and forwarding, NAT (Network Address Translation), and ACL (Access Control List) functions; it works deeply with smart network interface cards (NICs) to offload traffic classification and path execution functions to hardware, mainly undertaking the role of control and policy matching, significantly reducing CPU load; it supports elastic horizontal scaling, and can dynamically adjust processing capacity according to the scale of business traffic; it has a built-in multi-path management module that directly receives link selection policies issued by the control center.

[0092] 1.4 The Smart Network Interface Card (SmartNIC) has hardware acceleration capabilities, enabling: hardware offloading of Layer 3 / Layer 4 (L3 / L4) traffic processing logic, reducing CPU interrupts and improving forwarding performance; supporting traffic tagging by application and tenant dimensions, facilitating cross-cloud policy execution; and achieving zero-copy and batch processing optimization in conjunction with DPDK, adapting to large-scale enterprise traffic scenarios.

[0093] 1.5 Intelligent detector, responsible for link quality monitoring and data analysis: It adopts a hybrid monitoring mode that combines active detection (detection packet injection) and passive detection (service flow sampling); it works with TGW and smart NIC to achieve millisecond-level link anomaly switching; it predicts link degradation trends through AI (Artificial Intelligence) / ML (Machine Learning) algorithms and supports preventive scheduling.

[0094] 1.6 Cross-cloud backbone network, providing underlying transmission capabilities: supports multi-path routing, enabling dynamic routing based on feedback from intelligent detectors; provides unified abstract management across cloud vendors and multiple transmission protocols; supports differentiated services, allocating dedicated links for service flows with high SLA requirements.

[0095] 2 Key Processing Procedures 2.1 Traffic Identification and Classification A two-level identification mechanism is adopted: the smart network card performs fast matching based on the five-tuple (source Internet Protocol Address (IP), destination IP, source port, destination port, and protocol type); it identifies the service type through deep packet inspection (DPI) and pre-configured policies; and it outputs traffic classification results (normal forwarding flow, accelerated flow, probe flow, and fallback flow) and writes them into the flow table.

[0096] 2.2 Dynamic Path Selection and SLA Assurance Achieving quality-aware path optimization: Intelligent detectors periodically send probe messages to collect quality indicators such as latency, jitter, and packet loss rate; TGW calculates path scores based on a weighted scoring algorithm (SLA = α·latency + β·packet loss rate + γ·bandwidth); the control center dynamically adjusts routing strategies based on the scoring results to direct critical traffic to the optimal path; supports millisecond-level fast switching to ensure TCP / UDP session continuity.

[0097] 2.3 Intelligent Fault Tolerance and Rollback Provides a fault recovery mechanism: TGW monitors the operating status of smart network cards and detectors in real time; immediately reports any abnormalities to the control center; automatically switches to DPDK to ensure service continuity; and supports automatic switchback after fault recovery.

[0098] 2.4 Flow table unloading and acceleration To optimize forwarding performance: After completing the initial flow table calculation, the TGW sends the data to the Ternary Content-Addressable Memory (TCAM) of the smart NIC; subsequent packets are forwarded and encapsulated directly in the smart NIC; the TGW only handles abnormal flows and the first packet traffic.

[0099] 2.5 Cross-Domain Control and Management Provides unified management and control capabilities: The control center uniformly customizes SLA policies, routing policies, and isolation policies; these policies are distributed to each TGW node through standard APIs (Application Programming Interfaces) and control channels; a layered execution architecture is adopted: the TGW is responsible for policy parsing, and the smart NIC is responsible for hardware-accelerated execution; a quality feedback loop is established, and link quality data is collected regularly for policy optimization.

[0100] 3 Typical Application Examples A. Scene Configuration A company has deployed VPC clusters in East and South China (VPC-East, VPC-South) and needs to perform a full database synchronization. Cross-regional connectivity is achieved through a TGW cluster, providing three dedicated links: Link-1: 10 Gbps; Link-2: 5 Gbps; Link-3: 2 Gbps.

[0101] The control plane is configured with a total bandwidth pool (BWP) of 16 Gbps (expandable to 18 Gbps), a tenant base guaranteed bandwidth of 500 megabits per second (Mbps), and bursts up to 2 Gbps are allowed. The smart NIC's local token bucket is configured with a minimum rate of 200 Mbps and a burst capacity of 500 Mbps. The hot flow threshold is defined as: traffic ≥ 100 megabytes (MB) or ≥ 5000 packets per second (pps) within 5 seconds.

[0102] B. Operational Process During system initialization, the control plane distributes link topology and tenant policies to the smart NIC and TGW; normal service traffic is load balanced through five-tuple hashing and ACL checks; database synchronization triggers hot flow detection for sudden traffic bursts, and the smart NIC calculates path capacity locally and creates hot flow entries; when a link fails, the smart NIC reallocates traffic in real time and works with the TGW to ensure transmission continuity; after the traffic ends, resources are automatically reclaimed and billing information is reported.

[0103] This application achieves high-performance forwarding and intelligent path selection across cloud networks through the collaborative processing of smart NICs and TGWs, significantly improving service quality and resource utilization.

[0104] Figure 5 This is a block diagram illustrating an embodiment of a network traffic forwarding device provided in this application. Figure 5 As shown, the device includes: The acquisition module 51 is used to acquire the link quality information measured by the smart network card for cross-cloud links; The reporting module 52 is used to report the link quality information to the forwarding gateway in the cloud; The receiving module 53 is used to receive the path selection policy sent by the forwarding gateway, wherein the path selection policy is generated by the forwarding gateway based on the link quality information and preset policy information; The forwarding module 54 is used to forward network traffic according to the path selection strategy.

[0105] In one possible implementation, the acquisition module is specifically used for: At preset time intervals, link probe messages are generated according to a preset probe message format; The link probe message is sent to the peer forwarding gateway through the cross-cloud link; Receive the response message returned by the peer forwarding gateway; Based on the sending time of multiple link probe messages and the receiving time of the response messages, the link quality information of the cross-cloud link is determined. The link quality information includes at least one of transmission delay, delay jitter value, and packet loss rate.

[0106] In one possible implementation, the device further includes a monitoring module for: Monitor network data flow through the smart network card; If the network data flow rate is found to exceed a preset threshold, a burst flow event containing flow identification information is generated; The burst flow event is reported to the forwarding gateway, wherein the path selection strategy is generated by the forwarding gateway in response to the burst flow event.

[0107] In one possible implementation, the forwarding module is specifically used for: Based on the path selection strategy, determine the forwarding path of the data stream corresponding to the burst event; Based on the mapping relationship between the flow identifier information and the forwarding path, a flow table entry is created in the hardware forwarding table of the smart network card; For the received network traffic, the packet characteristics of the network traffic are matched with the flow table entries in the hardware forwarding table; If a match is successful, the network traffic is forwarded on the smart network interface card according to the forwarding path specified in the matched flow table entry. If the matching fails, the flow identification information corresponding to the network traffic is reported to the forwarding gateway, and the forwarding gateway receives the forwarding instruction issued by the forwarding gateway based on the flow identification information, and forwards the network traffic according to the forwarding instruction.

[0108] In one possible implementation, the device further includes a switching module for: The system receives a path switching instruction from the forwarding gateway. The path switching instruction is used to indicate that the forwarding path of the target data stream be switched to a new path. The path switching instruction includes the flow identification information of the target data stream and the corresponding new path. Update the forwarding path information in the local policy table corresponding to the flow identifier information to the new path; For subsequent network traffic that belongs to the target data stream, the network traffic is forwarded according to the new path updated in the local policy table.

[0109] In one possible implementation, the apparatus further includes a replication module for: Within a preset switching time window, each network traffic packet belonging to the target data stream is copied and forwarded via the original path and the new path respectively. After the switching time window ends, network traffic forwarding the target data stream to the original path is stopped.

[0110] In one possible implementation, the path switching instruction is further configured to instruct session state synchronization of the target data stream; the apparatus further includes a processing module configured to: In response to the path switching command, communicate with the forwarding gateway to obtain the current session state information of the target data stream; During the forwarding of network traffic according to the new path updated in the local policy table, the network traffic is processed based on the current session state information to ensure that the target data stream session state is normal.

[0111] In one possible implementation, the apparatus further includes a transmitting module for: Monitor the hardware acceleration function status of the smart network card; When the hardware acceleration function is detected to be in an abnormal state, a function abnormality notification is sent to the forwarding gateway. Stop executing the step of forwarding network traffic according to the path selection strategy, and upload the received network traffic to the forwarding gateway; In response to the functional error notification, the forwarding gateway switches to software forwarding mode to process the uploaded network traffic.

[0112] In one possible implementation, the link quality information includes multiple quality parameters, and the preset policy information includes tenant service level agreement requirements. The path selection strategy is generated by the forwarding gateway by performing the following operations: The weight parameters corresponding to each of the quality parameters are determined according to the requirements of the tenant service level agreement. The quality parameters are weighted according to the weight parameters corresponding to each quality parameter to obtain the comprehensive score of each optional path; The path with the highest overall score is selected from the available paths as the designated forwarding path, and the path selection strategy is generated based on the designated forwarding path.

[0113] In one possible implementation, the forwarding module is specifically used for: Identify the service category corresponding to the network traffic; The network traffic is forwarded according to the path selection strategy and the service category.

[0114] like Figure 6 As shown in the figure, this application provides a device including a processor 111, a communication interface 112, a memory 113, and a communication bus 114, wherein the processor 111, the communication interface 112, and the memory 113 communicate with each other through the communication bus 114. Memory 113 is used to store computer programs; In one embodiment of this application, the processor 111, when executing a program stored in the memory 113, implements the network traffic forwarding method provided in any of the foregoing method embodiments, applied to a smart network interface card, including: Obtain the link quality information measured by the smart network interface card for cross-cloud links; The link quality information is reported to the forwarding gateway in the cloud. The system receives a path selection policy from the forwarding gateway, wherein the path selection policy is generated by the forwarding gateway based on the link quality information and preset policy information. Network traffic is forwarded according to the path selection strategy described above.

[0115] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the network traffic forwarding method provided in any of the foregoing method embodiments.

[0116] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0117] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0118] It should be understood that the terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “described” as used herein may also include the plural forms. The terms “comprising,” “including,” “containing,” and “having” are inclusive and therefore indicate the presence of the stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not construed as requiring them to be performed in a particular order described or illustrated unless the order of performance is explicitly indicated. It should also be understood that additional or alternative steps may be used.

[0119] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A network traffic forwarding method, characterized in that, Applied to smart network interface cards, the method includes: Obtain the link quality information measured by the smart network interface card for cross-cloud links; The link quality information is reported to the forwarding gateway in the cloud. The system receives a path selection policy from the forwarding gateway, wherein the path selection policy is generated by the forwarding gateway based on the link quality information and preset policy information. Network traffic is forwarded according to the path selection strategy described above.

2. The method according to claim 1, characterized in that, The step of obtaining the link quality information measured by the smart network interface card for cross-cloud links includes: At preset time intervals, link probe messages are generated according to a preset probe message format; The link probe message is sent to the peer forwarding gateway through the cross-cloud link; Receive the response message returned by the peer forwarding gateway; Based on the sending time of multiple link probe messages and the receiving time of the response messages, the link quality information of the cross-cloud link is determined. The link quality information includes at least one of transmission delay, delay jitter value, and packet loss rate.

3. The method according to claim 1, characterized in that, The method further includes: Monitor network data flow through the smart network card; If the network data flow rate is found to exceed a preset threshold, a burst flow event containing flow identification information is generated; The burst flow event is reported to the forwarding gateway, wherein the path selection strategy is generated by the forwarding gateway in response to the burst flow event.

4. The method according to claim 3, characterized in that, The forwarding of network traffic according to the path selection strategy includes: Based on the path selection strategy, determine the forwarding path of the data stream corresponding to the burst event; Based on the mapping relationship between the flow identifier information and the forwarding path, a flow table entry is created in the hardware forwarding table of the smart network card; For the received network traffic, the packet characteristics of the network traffic are matched with the flow table entries in the hardware forwarding table; If a match is successful, the network traffic is forwarded on the smart network interface card according to the forwarding path specified in the matched flow table entry. If the matching fails, the flow identification information corresponding to the network traffic is reported to the forwarding gateway, and the forwarding gateway receives the forwarding instruction issued by the forwarding gateway based on the flow identification information, and forwards the network traffic according to the forwarding instruction.

5. The method according to claim 1, characterized in that, The method further includes: The system receives a path switching instruction from the forwarding gateway. The path switching instruction is used to indicate that the forwarding path of the target data stream be switched to a new path. The path switching instruction includes the flow identification information of the target data stream and the corresponding new path. Update the forwarding path information in the local policy table corresponding to the flow identifier information to the new path; For subsequent network traffic that belongs to the target data stream, the network traffic is forwarded according to the new path updated in the local policy table.

6. The method according to claim 5, characterized in that, After updating the forwarding path information corresponding to the flow identifier information in the local policy table to the new path, the method further includes: Within a preset switching time window, each network traffic packet belonging to the target data stream is copied and forwarded via the original path and the new path respectively. After the switching time window ends, network traffic forwarding the target data stream to the original path is stopped.

7. The method according to claim 5, characterized in that, The path switching instruction is also used to instruct session state synchronization of the target data stream; The method further includes: In response to the path switching command, communicate with the forwarding gateway to obtain the current session state information of the target data stream; During the forwarding of network traffic according to the new path updated in the local policy table, the network traffic is processed based on the current session state information to ensure that the target data stream session state is normal.

8. The method according to claim 1, characterized in that, The method further includes: Monitor the hardware acceleration function status of the smart network card; When the hardware acceleration function is detected to be in an abnormal state, a function abnormality notification is sent to the forwarding gateway. Stop executing the step of forwarding network traffic according to the path selection strategy, and upload the received network traffic to the forwarding gateway; In response to the functional error notification, the forwarding gateway switches to software forwarding mode to process the uploaded network traffic.

9. The method according to claim 1, characterized in that, The link quality information includes multiple quality parameters, and the preset policy information includes tenant service level agreement requirements. The path selection strategy is generated by the forwarding gateway by performing the following operations: The weight parameters corresponding to each of the quality parameters are determined according to the requirements of the tenant service level agreement. The quality parameters are weighted according to the weight parameters corresponding to each quality parameter to obtain the comprehensive score of each optional path; The path with the highest overall score is selected from the available paths as the designated forwarding path, and the path selection strategy is generated based on the designated forwarding path.

10. The method according to claim 1, characterized in that, The forwarding of network traffic according to the path selection strategy includes: Identify the service category corresponding to the network traffic; The network traffic is forwarded according to the path selection strategy and the service category.

11. A network traffic forwarding device, characterized in that, The device, applied to a smart network interface card, includes: The acquisition module is used to acquire the link quality information measured by the smart network card for cross-cloud links; The reporting module is used to report the link quality information to the forwarding gateway in the cloud; The receiving module is used to receive the path selection policy issued by the forwarding gateway, wherein the path selection policy is generated by the forwarding gateway based on the link quality information and preset policy information; The forwarding module is used to forward network traffic according to the path selection strategy.

12. An electronic device, characterized in that, include: A processor and a memory, the processor being configured to execute a network traffic forwarding program stored in the memory to implement the network traffic forwarding method according to any one of claims 1-10.

13. A storage medium, characterized in that, The storage medium stores one or more programs, which can be executed by one or more processors to implement the network traffic forwarding method according to any one of claims 1-10.

Images