Multi-protocol fusion packet transmission adaptation method and switch
By using a multi-protocol converged switch that identifies business scenarios, drives policies, and dynamically allocates resources, the problem of insufficient resource allocation in existing technologies has been solved, achieving differentiated performance guarantees and protocol scalability, and improving resource utilization and adaptability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ANHUI SHUHANG TECHNOLOGY CO LTD
- Filing Date
- 2026-02-03
- Publication Date
- 2026-05-15
AI Technical Summary
Existing multi-protocol converged switches struggle to achieve refined and differentiated resource allocation and performance assurance in complex production environments, leading to performance jitter and packet loss in critical business traffic, and lacking flexible protocol scalability.
By identifying business scenarios and mapping policies, policy-driven packet processing and dynamic resource adaptation, combined with software-defined network controllers and programmable ASIC chips, differentiated forwarding policies and resource allocation are achieved, supporting lossless Ethernet features and unknown protocol parsing.
It enables differentiated performance guarantees for different services on a shared network, improves resource utilization and flexibility, and adapts to complex traffic demands.
Smart Images
Figure CN122053526A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of switch technology, specifically to a multi-protocol convergence packet transmission adaptation method and a switch. Background Technology
[0002] In recent years, with the rapid development of cloud computing, big data, and artificial intelligence technologies, data center network architecture is undergoing profound changes. Traditionally, different business workloads (such as data communication, storage access, and high-performance computing) typically rely on independent, heterogeneous networks (such as Ethernet, Fibre Channel, and InfiniBand), resulting in complex infrastructure, high costs, and difficult operation and maintenance. To address this issue, multi-protocol converged packet transport switches have emerged. These switches are designed to carry various heterogeneous network traffic over a unified IP / Ethernet physical network through advanced tunneling encapsulation technologies (such as VXLAN and NVGRE), high-performance hardware, and software-defined networking (SDN) capabilities, achieving "one network for multiple uses" and significantly improving resource utilization, flexibility, and scalability.
[0003] However, deploying multi-protocol converged switches in complex and ever-changing real-world production environments still presents significant adaptability challenges. Existing solutions often employ static or coarse-grained configuration strategies, making it difficult to provide fine-grained and differentiated resource allocation and performance assurance for different business scenarios (such as virtualization migration, storage network convergence, and RDMA applications). Specifically, in converged traffic environments, critical business traffic (such as storage and RDMA) may suffer from performance jitter and packet loss due to a lack of independent buffer management, precise QoS policies, and dynamic congestion control. Simultaneously, the network lacks programmable and flexible processing capabilities when facing future unknown protocols or customized business requirements, resulting in insufficient protocol scalability.
[0004] To address the aforementioned issues, we propose an improvement: a multi-protocol fusion packet transmission adaptation method and switch. Summary of the Invention
[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a multi-protocol converged packet transmission adaptation method and switch, comprising the following steps: S1: Business scenario identification and policy mapping, identifying the preset business scenario type to which the business data stream input at the ingress port belongs, and configuring differentiated forwarding processing policies for different types of business scenarios; the business scenario type includes at least two of the following: virtualization network scenario, storage network convergence scenario, and high-performance computing network convergence scenario; S2: Policy-driven packet processing, which performs corresponding tunnel encapsulation, priority marking, queue scheduling and congestion control operations on the service data stream according to the forwarding processing policy mapped to the service data stream; S3: Dynamic resource adaptation, based on real-time network status monitoring data, dynamically adjusts the hardware resource allocation parameters associated with each of the forwarding processing strategies, including queue buffers and forwarding bandwidth.
[0006] As a preferred technical solution of the present invention, in step S1, the method of identifying the business scenario type is as follows: parsing the packet header information of the business data stream, and classifying and mapping the data stream to the corresponding business scenario type and forwarding processing strategy according to the protocol type, target port number or preset deep packet inspection rules.
[0007] As a preferred technical solution of the present invention, in step S2, the forwarding processing strategy is uniformly distributed and orchestrated by the software-defined network controller, and the switch receives and executes the strategy through the southbound interface protocol.
[0008] As a preferred technical solution of the present invention, the forwarding processing strategy configured for the storage network convergence scenario or the high-performance computing network convergence scenario enables lossless Ethernet features, which include priority-based flow control and explicit congestion notification.
[0009] As a preferred technical solution of the present invention, it also includes performance baseline establishment and verification. During the period of stable business traffic, the baseline values of key performance indicators under each business scenario type are measured and recorded. The key performance indicators include latency, jitter and throughput. During the operation period, the baseline values are continuously monitored and compared with the baseline values to trigger the re-adaptation of policies or resources.
[0010] As a preferred embodiment of the present invention, the method is implemented by user-defined processing logic deployed in the programmable ASIC chip of the switch, which supports parsing and adapting forwarding of unknown or customized protocol messages.
[0011] A multi-protocol converged packet transmission switch, comprising: At least one programmable ASIC chip for line-speed processing of packet data; The strategy execution unit, integrated into the programmable ASIC chip, is used to store and execute differentiated forwarding processing strategies associated with different business scenario types; The dynamic resource management unit is used to monitor port and queue status, and dynamically allocate queue buffers and scheduling bandwidth according to the instructions of the policy execution unit and monitoring data.
[0012] As a preferred embodiment of the present invention, the programmable ASIC chip supports the P4 programming language, the physical ports of the switch have deep packet buffering capability, and support at least one tunnel encapsulation protocol among VXLAN, NVGRE, and GENEVE, as well as at least one converged protocol among RoCE and FCoE.
[0013] The beneficial effects of this invention are: a multi-protocol converged packet transmission adaptation method and switch, which organically combines the three links of service identification, policy execution and dynamic resource allocation, enables the switch to intelligently respond to complex and ever-changing converged traffic demands, provide differentiated and guaranteed performance isolation for different services on a shared physical network, and improve overall resource utilization. Attached Figure Description
[0014] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a multi-protocol fusion packet transmission adaptation method according to the present invention. Detailed Implementation
[0015] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.
[0016] Example: Figure 1 As shown, a multi-protocol converged packet transmission adaptation method and switch include the following steps: S1: Business Scenario Identification and Policy Mapping. Identify the preset business scenario type to which the business data stream input at the ingress port belongs, and configure differentiated forwarding and processing policies for different types of business scenarios. The business scenario type includes at least two of the following: virtualization network scenario, storage network convergence scenario, and high-performance computing network convergence scenario. The business scenario type is a label for a set of predefined services with common network service quality requirements. Virtualized network scenarios are characterized by the extensive use of overlay network protocols such as VXLAN for virtual machine migration and east-west communication, and require high levels of network isolation and resilience. Storage network convergence scenarios are characterized by carrying protocols such as FCoE or iSCSI, and are extremely sensitive to packet loss (requiring zero) and stable latency; High-performance computing network scenarios are characterized by carrying the RoCE protocol, requiring ultra-low latency and high throughput; S2: Policy-driven packet processing. Based on the forwarding processing policy mapped to the service data flow, it performs corresponding tunnel encapsulation, priority marking, queue scheduling, and congestion control operations on the service data flow. Based on the policy mapped by S1, the switch's data plane performs a series of specific and differentiated operations on the data packets. S3: Dynamic resource adaptation, based on real-time network status monitoring data, dynamically adjusts the hardware resource allocation parameters associated with each forwarding processing strategy. The hardware resource allocation parameters include queue buffer and forwarding bandwidth. The monitoring basis includes, but is not limited to, the current depth of each queue, the real-time port utilization, the number of ECN tags, and the triggering frequency of PFC pause frames. Dynamic adjustments, which can be automatically executed based on monitoring data, can temporarily increase buffer quotas for critical business queues experiencing congestion; when links are idle, the bandwidth ratio in the weighted scheduling of services with increasing bandwidth demand can be dynamically adjusted; this step transforms resource allocation from static configuration to dynamic response, which is the key to achieving lossless integration and efficient utilization.
[0017] Furthermore, in step S1, the method for identifying the business scenario type is as follows: parse the header information of the business data stream, and classify and map the data stream to the corresponding business scenario type and forwarding processing strategy according to the protocol type, target port number or preset deep packet inspection rules.
[0018] Furthermore, in step S2, the forwarding processing policy is uniformly distributed and orchestrated by the software-defined network controller, and the switch receives and executes the policy through the southbound interface protocol.
[0019] Preferably, the forwarding processing strategy configured for storage network convergence scenarios or high-performance computing network convergence scenarios enables lossless Ethernet features, including priority-based flow control and explicit congestion notification.
[0020] Furthermore, it also includes the establishment and verification of performance baselines. During periods of stable business traffic, the baseline values of key performance indicators for each business scenario type are measured and recorded. Key performance indicators include latency, jitter, and throughput. During operation, the baseline values are continuously monitored and compared with the baseline values to trigger the re-adaptation of policies or resources.
[0021] Preferably, the method is implemented through user-defined processing logic deployed in the programmable ASIC chip of the switch, which supports parsing and adapting forwarding of unknown or customized protocol messages.
[0022] A multi-protocol converged packet transmission switch, comprising: At least one programmable ASIC chip for line-speed processing of packet data; The strategy execution unit, integrated into a programmable ASIC chip, is used to store and execute differentiated forwarding processing strategies associated with different business scenario types; Includes: a policy table / flow table, which stores flow classification rules issued by the controller or configured locally and the corresponding action set (i.e., the mapping result of S1 and the execution instruction of S2); and a matching-action pipeline, which searches and matches in the policy table based on the characteristics of the incoming packet and executes a series of associated actions in sequence (such as adding headers, modifying fields, and sending to the queue). The dynamic resource management unit is used to monitor port and queue status, and dynamically allocate queue buffers and scheduling bandwidth according to the instructions of the policy execution unit and monitoring data. This includes: a monitoring counter that collects real-time status information such as queue length and port traffic; and a resource scheduler that dynamically allocates and adjusts buffer and bandwidth resources among different queues and ports according to the requests of the execution unit based on preset algorithms and policies.
[0023] The programmable ASIC chip supports the P4 programming language, the physical ports of the switch have deep packet buffering capabilities, and support at least one tunnel encapsulation protocol among VXLAN, NVGRE, and GENEVE, as well as at least one converged protocol among RoCE and FCoE.
[0024] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A multi-protocol fusion packet transmission adaptation method, characterized in that, Includes the following steps: S1: Business scenario identification and policy mapping, identifying the preset business scenario type to which the business data stream input at the ingress port belongs, and configuring differentiated forwarding processing policies for different types of business scenarios; the business scenario type includes at least two of the following: virtualization network scenario, storage network convergence scenario, and high-performance computing network convergence scenario; S2: Policy-driven packet processing, which performs corresponding tunnel encapsulation, priority marking, queue scheduling and congestion control operations on the service data stream according to the forwarding processing policy mapped to the service data stream; S3: Dynamic resource adaptation, based on real-time network status monitoring data, dynamically adjusts the hardware resource allocation parameters associated with each of the forwarding processing strategies, including queue buffers and forwarding bandwidth.
2. The multi-protocol fusion packet transmission adaptation method according to claim 1, characterized in that, In step S1, the method for identifying the business scenario type is as follows: parse the header information of the business data stream, and classify and map the data stream to the corresponding business scenario type and forwarding processing strategy according to the protocol type, target port number or preset deep packet inspection rules.
3. The multi-protocol fusion packet transmission adaptation method according to claim 1, characterized in that, In step S2, the forwarding processing policy is uniformly distributed and orchestrated by the software-defined network controller, and the switch receives and executes the policy through the southbound interface protocol.
4. The multi-protocol fusion packet transmission adaptation method according to claim 1, characterized in that, For the forwarding processing strategy configured for the aforementioned storage network convergence scenario or high-performance computing network convergence scenario, lossless Ethernet features are enabled, including priority-based flow control and explicit congestion notification.
5. The multi-protocol fusion packet transmission adaptation method according to claim 1, characterized in that, It also includes the establishment and verification of performance baselines. During periods of stable business traffic, benchmark values of key performance indicators for each business scenario type are measured and recorded. These key performance indicators include latency, jitter, and throughput. During operation, the benchmark values are continuously monitored and compared with the benchmark values to trigger the re-adaptation of policies or resources.
6. The packet transmission adaptation method with multi-protocol fusion according to claim 1, characterized in that, The method is implemented through user-defined processing logic deployed in the programmable ASIC chip of the switch, which supports parsing and adapting forwarding of unknown or customized protocol messages.
7. A multi-protocol converged packet transmission switch according to claims 1-6, characterized in that, include: At least one programmable ASIC chip for line-speed processing of packet data; The strategy execution unit, integrated into the programmable ASIC chip, is used to store and execute differentiated forwarding processing strategies associated with different business scenario types; The dynamic resource management unit is used to monitor port and queue status, and dynamically allocate queue buffers and scheduling bandwidth according to the instructions of the policy execution unit and monitoring data.
8. A multi-protocol converged packet transmission switch according to claim 7, characterized in that, The programmable ASIC chip supports the P4 programming language, the physical ports of the switch have deep packet buffering capability, and support at least one tunnel encapsulation protocol among VXLAN, NVGRE, and GENEVE, as well as at least one converged protocol among RoCE and FCoE.