A flow control method, device, equipment and medium

Abstract

This invention discloses a traffic control method, apparatus, device, and medium, relating to the field of optoelectronic hybrid switching technology. The method includes: classifying and parsing access data services according to flow type and storing them in an accumulated buffer queue; pre-scheduling outgoing data using a differentiated scheduling strategy when the queue data volume reaches a preset threshold, encapsulating the outgoing data into fixed-length cells and storing them in an egress buffer queue; initiating a bandwidth request to the destination processing end when the remaining bandwidth cannot meet service demands; updating resource configuration after receiving bandwidth authorization, and controlling the forwarding of cells in the queue via optical switching units or electrical packet switching units based on optical and electrical gating states; and completing cell decapsulation, data flow reassembly, and egress buffer scheduling output after the cells arrive at the destination processing end. This allows for dynamic and flexible allocation of optoelectronic switching bandwidth resources, supports the splitting and carrying of service flows between optoelectronic switching paths, improves bandwidth utilization, saves power consumption, and ensures service transmission stability.

Description

Technical Field

[0001] This invention relates to the field of optoelectronic hybrid switching technology, and in particular to a flow control method, apparatus, equipment and medium. Background Technology

[0002] As the scale of AI model training continues to expand and GPU / TPU clusters grow, traditional Electronic Packet Switching (EPS) architecture suffers from bandwidth bottlenecks, high power consumption, and high expansion costs, making it difficult to adapt to the needs of high-bandwidth service transmission. Optical Circuit Switching (OCS) offers advantages such as high throughput, low power consumption, and easy smooth evolution, but its one-to-one port circuit connection characteristic limits its flexibility in independently scheduling services. While hybrid optoelectronic switches combine the technological advantages of both EPS and OCS, they still have significant shortcomings: traditional hybrid optoelectronic switching equipment handles traffic management for electronic packet switching and optical switching independently, resulting in low bandwidth utilization, complex structure, high resource consumption, poor flexibility and overall transmission performance, and difficulty in balancing resource and power consumption optimization with service transmission stability. Summary of the Invention

[0003] The purpose of this invention is to provide a traffic management method, apparatus, device, and medium that can dynamically and flexibly allocate optoelectronic switching bandwidth resources, support the splitting and carrying of service flows between optoelectronic switching paths, improve bandwidth utilization, save system resources and operating power consumption, and ensure the stability of service transmission.

[0004] To address the aforementioned technical problems, this invention provides a flow control method applied to a hybrid optoelectronic switching system, comprising: The incoming data services are stream parsed and classified, and stored in the cumulative cache queue according to the service flow type; When the amount of data in the cumulative buffer queue reaches the preset message accumulation threshold, pre-schedule dequeueing is performed according to the differentiated scheduling strategy, and the dequeueed data is encapsulated into fixed-length information cells and stored in the exit buffer queue. Maintain the optoelectronic bandwidth gating list and initiate a bandwidth request to the destination processing terminal when the remaining bandwidth does not meet the service carrying requirements; After receiving the bandwidth authorization from the destination processing end, the bandwidth resources are updated, and the cells of the egress buffer queue are forwarded via the electrical packet switching unit according to the electrical gating state, and the cells of the egress buffer queue are forwarded via the optical switching unit according to the optical gating state. After the information cell is forwarded to the destination processing end, the information cell is decapsulated and the data stream is reassembled, and then output after being scheduled by the service-side egress cache.

[0005] To address the aforementioned technical problems, the present invention also provides a flow control device applied to a hybrid optoelectronic switching system, comprising: The incoming service processing unit is used to perform flow parsing and classification on the incoming data services and store them in the cumulative cache queue according to the service flow type. When the data volume of the cumulative cache queue reaches the preset message accumulation threshold, it performs pre-scheduling and dequeueing according to the differentiated scheduling strategy, encapsulates the dequeueed data into fixed-length cells, and stores them in the outgoing cache queue. It maintains the optoelectronic bandwidth gating list and initiates a bandwidth application to the destination processing end when the remaining bandwidth does not meet the service carrying requirements. After receiving the bandwidth authorization issued by the destination processing end, it updates the bandwidth resources and controls the cells in the outgoing cache queue to be forwarded through the electrical packet switching unit according to the electrical gating state, and controls the cells in the outgoing cache queue to be forwarded through the optical switching unit according to the optical gating state. The electrical packet switching unit is used to receive the cells sent by the incoming service processing unit and to forward the cells in a statistical multiplexing manner. The optical switching unit is used to receive the cells sent by the incoming service processing unit and forward the cells in a time-division exclusive mode. The outbound service processing unit is used to receive cells that arrive after being forwarded by the electrical packet switching unit and the optical switching unit, decapsulate and reassemble the data streams of the cells, and output them outward after being scheduled by the service-side egress buffer; it is also used to receive bandwidth requests from the inbound service processing unit and issue bandwidth authorizations; the outbound service processing unit includes multiple processing ends, and the destination processing end is any one of the processing ends in the outbound service processing unit.

[0006] To address the aforementioned technical problems, the present invention also provides an electronic device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the aforementioned traffic control method.

[0007] To address the aforementioned technical problems, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the aforementioned traffic control method.

[0008] The beneficial effects of this invention are as follows: The traffic control method provided by this invention classifies access data services by stream parsing and storing them in an accumulated buffer queue according to the service flow type. When the data volume in the accumulated buffer queue reaches a preset message accumulation threshold, pre-scheduling and dequeuing are performed according to a differentiated scheduling strategy. The dequeued data is encapsulated into fixed-length cells and stored in the egress buffer queue. By maintaining an optoelectronic bandwidth gating list, when bandwidth resources are insufficient, a bandwidth request is initiated to the destination processing end. After receiving bandwidth authorization, the bandwidth resources are updated, and the cells in the egress buffer queue are forwarded via the electrical packet switching unit according to the electrical gating state, and the cells in the egress buffer queue are forwarded via the optical switching unit according to the optical gating state. This enables dynamic and flexible allocation of optoelectronic switching bandwidth resources, significantly improving the overall bandwidth utilization. This scheme uniformly encapsulates optoelectronic switching services into cell slices, supporting flexible load sharing between the same service flow or different service flows on the electrical packet switching and optical switching paths, effectively saving system resources and operating power consumption. In addition, after the cells are forwarded to the destination processing end, unified decapsulation, data flow reassembly, and egress buffer scheduling output further ensure the stability of service transmission.

[0009] In addition, the present invention also provides a corresponding traffic control device, electronic device and computer-readable storage medium for the traffic control method, which have the same or corresponding technical features as the traffic control method mentioned above, and have the same effect. Attached Figure Description

[0010] To more clearly illustrate the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A flowchart of the traffic control method provided in the embodiments of the present invention; Figure 2 A block diagram of the flow control mechanism of the optoelectronic hybrid switching system provided in an embodiment of the present invention; Figure 3 This is an internal structural diagram of the incoming service processing unit provided in an embodiment of the present invention; Figure 4 This is a bandwidth request and authorization path diagram provided in an embodiment of the present invention; Figure 5 A block diagram illustrating the principle of optoelectronic hybrid bandwidth application and authorization provided in this embodiment of the invention; Figure 6 A flowchart of photoelectric hybrid flow control provided in an embodiment of the present invention; Figure 7 The encapsulation strategy for latency-sensitive services provided in the embodiments of the present invention; Figure 8 This is a timing diagram of the latency-sensitive service bandwidth authorization mechanism provided in an embodiment of the present invention; Figure 9 A schematic diagram illustrating the encapsulation strategy for elephant stream services provided in an embodiment of the present invention; Figure 10 A schematic diagram illustrating the bandwidth authorization mechanism decision-making for elephant stream services provided in this embodiment of the invention; Figure 11 A timing diagram of the bandwidth authorization mechanism for the Elephant Stream service provided in this embodiment of the invention; Figure 12 A timing diagram of the hybrid mode bandwidth licensing mechanism provided in an embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of the present invention.

[0013] It should be noted that, in the description of this invention, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. The terms "first," "second," etc., used in this invention are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0014] To enable those skilled in the art to better understand the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0015] The specific application environment architecture or specific hardware architecture on which the execution of the traffic control method depends is described here.

[0016] The embodiments of the present invention provide a traffic control method, and the method is described in detail in conjunction with the execution flow of the traffic control method. Figure 1 A flowchart of the traffic control method provided in the embodiments of the present invention is shown below. Figure 1 As shown, this method is applied to a hybrid optoelectronic switching system, including: S101. Perform flow parsing and classification on the accessed data services, and store them in the cumulative cache queue according to the service flow type.

[0017] Figure 2 A block diagram illustrating the traffic control mechanism of a hybrid optoelectronic switching system provided in an embodiment of the present invention. Figure 2 As shown, the optoelectronic hybrid switching system of the present invention may include an Ingress Management (IM) unit, an Electrical Packet Switching (EPS) unit, an Optical Switching (OCS) unit, and an Output Management (OM) unit. In an n×n structure optoelectronic hybrid switching system, there are n Ingress Management (IM) units, n Output Management (OM) units, 1 Electrical Packet Switching (EPS) unit, and 1 Optical Switching (OCS) unit.

[0018] Step S101 involves the Incoming Service Processing Unit (IM) performing flow parsing and classification on the incoming data services. This accurately distinguishes service flows with different attributes and requirements, and then stores them in the corresponding cumulative cache queues according to the service flow type, enabling isolated control of inbound traffic. The cumulative cache queue here is a cache queue used for temporarily caching service data flows. This cumulative cache queue can be called the Virtual Output Queue (VOQ) cumulative cache queue.

[0019] S102. When the amount of data in the cumulative buffer queue reaches the preset message accumulation threshold, pre-schedule the dequeue according to the differentiated scheduling strategy, encapsulate the dequeue data into fixed-length cells and store them in the exit buffer queue.

[0020] Step S102 involves the inbound service processing unit (IM) performing pre-scheduled dequeue processing according to a differentiated scheduling strategy when the data volume in the VOQ cumulative buffer queue reaches a preset message accumulation threshold. The dequeueed data is then uniformly encapsulated into fixed-length cells and stored in the outbound buffer queue. The outbound buffer queue can be referred to as the Virtual Output Module (VOM) outbound buffer queue. It should be noted that the inbound service processing unit (IM) can include both the virtual outbound cumulative buffer queue (i.e., the VOQ cumulative buffer queue) and the Virtual Output Module (VOM) outbound buffer queue.

[0021] S103. Maintain the optoelectronic bandwidth gating list and initiate a bandwidth request to the destination processing terminal when the remaining bandwidth does not meet the service carrying requirements.

[0022] It should be noted that the optoelectronic bandwidth gating list is a management list used to record the bandwidth resource usage and remaining bandwidth of electrical and optical paths. The optoelectronic bandwidth gating list includes the optical bandwidth gating list and the electrical bandwidth gating list.

[0023] Step S103 involves the Incoming Service Processing Unit (IM) maintaining a real-time optoelectronic bandwidth gating list, recording and monitoring the bandwidth occupancy, remaining resources, and gating status of optical and electrical paths, and dynamically sensing the current bandwidth requirements of the service. The electrical path is the logical transmission path through which cells are stored, forwarded, parsed, and switched via the Electrical Packet Switching Unit (EPS). The optical path is the logical transmission path through which cells are directly switched via the Optical Switching Unit (OCS). When it is detected that the remaining bandwidth cannot match the service transmission requirements, a bandwidth request is actively initiated to the destination processing unit. The destination processing unit is one of the multiple Outgoing Service Processing Units (OMs).

[0024] S104. After receiving the bandwidth authorization from the destination processing end, update the bandwidth resources, and control the cells in the egress buffer queue to be forwarded via the electrical packet switching unit according to the electrical gating state, and control the cells in the egress buffer queue to be forwarded via the optical switching unit according to the optical gating state.

[0025] It should be noted that electrical gating status refers to the switching control status in the electrical bandwidth gating list that indicates whether the current electrical path has available bandwidth, used to determine whether the corresponding cell is allowed to leave the queue and be forwarded by the electrical packet switching unit. Optical gating status refers to the switching control status in the optical bandwidth gating list that indicates whether the current optical path has been allocated time slot resources, used to determine whether the corresponding cell is allowed to leave the queue and be forwarded by the optical switching unit.

[0026] Step S104 involves the Incoming Service Processing Unit (IM) receiving the bandwidth authorization from the destination processing unit, updating the bandwidth resources, and controlling the forwarding of cells in the queue via either the optical switching unit (OSU) or the electrical packet switching unit (OCS) based on the optical gating state and electrical gating state. After receiving cells from the queue, the EPS can perform packet switching and forwarding of the corresponding cells according to a statistical usage method. Similarly, the OCS can perform optical path switching and forwarding of the corresponding cells according to a time-division exclusive usage method.

[0027] S105. After the information cell is forwarded to the destination processing end, the information cell is decapsulated and the data stream is reassembled, and then output after being scheduled by the service side egress buffer.

[0028] In the traffic control method provided by this invention, access data services are classified and stored in an accumulated buffer queue according to the service flow type. When the data volume in the accumulated buffer queue reaches a preset message accumulation threshold, pre-scheduling and dequeuing are performed according to a differentiated scheduling strategy. The dequeued data is encapsulated into fixed-length cells and stored in the egress buffer queue. By maintaining an optoelectronic bandwidth gating list, a bandwidth request is initiated to the destination processing end when bandwidth resources are insufficient. After receiving bandwidth authorization, the bandwidth resources are updated, and the cells in the egress buffer queue are forwarded via the electrical packet switching unit (EPS) according to the electrical gating state, and the cells in the egress buffer queue are forwarded via the optical switching unit (OCS) according to the optical gating state. This allows for dynamic and flexible allocation of optoelectronic switching bandwidth resources, significantly improving the overall bandwidth utilization. This scheme uniformly encapsulates optoelectronic switching services into cell slices, supporting flexible load sharing between the same service flow or different service flows on the electrical packet switching and optical switching paths, effectively saving system resources and operating power consumption. Furthermore, after the cells are forwarded to the destination processing end, unified decapsulation, data flow reassembly, and egress buffer scheduling output further ensure the stability of service transmission.

[0029] Furthermore, in a specific implementation, in the traffic control method provided in the embodiments of the present invention, step S101 performs flow parsing and classification on the accessed data services and stores them in the cumulative cache queue according to the service flow type. Specifically, it may include: first, receiving externally accessed data services and reading the packet-related fields and traffic information in the data services; then, performing flow parsing and service classification on the read packet-related fields and traffic information, identifying service flow types with different priorities; and storing the data services in the corresponding priority cumulative cache queue according to the priority of the service flow type.

[0030] Figure 3 This is an internal structural diagram of the incoming service processing unit provided in an embodiment of the present invention. Figure 3 As shown, the Incoming Service Processing Unit (IM) may include a VOQ cumulative cache module. In implementation, after data services are input, the VOQ cumulative cache module can perform flow parsing and property classification based on message-related fields and traffic information, identify service flow types with different priorities, and store service packets into the corresponding priority VOQ cumulative cache queue according to flow type. This achieves accurate identification, hierarchical isolation, and orderly caching of incoming service flows.

[0031] Furthermore, in a specific implementation, in the traffic control method provided in the embodiments of the present invention, step S102 performs pre-scheduled dequeueing according to a differentiated scheduling strategy, and encapsulates the dequeueing data into fixed-length cells and stores them in the egress cache queue. Specifically, it may include: first, performing pre-scheduled dequeueing from the cumulative cache queue according to the priority of the cumulative cache queue, the cumulative number of queue messages, and the downstream egress queue level information, using a differentiated scheduling strategy; then, encapsulating the dequeueing service data to generate fixed-length cells carrying sequence numbers, service flow types, and forwarding tables; finally, storing the fixed-length cells in the electrical path queue or optical path queue in the egress cache queue according to a preset diversion rule that includes traffic size attributes, queue load status, or path selection rules.

[0032] like Figure 3 As shown, the Incoming Service Processing Unit (IM) may also include a pre-scheduling and slice encapsulation module, a VOM egress buffer module, and an optoelectronic hybrid scheduling module. Specifically, when the data volume in the VOQ cumulative buffer queue reaches a preset packet accumulation threshold, the pre-scheduling and slice encapsulation module can perform a pre-scheduling dequeue operation. Through the packet accumulation mechanism, the system does not need to schedule every clock cycle, significantly reducing the scheduling processing frequency, circuit timing pressure, and improving system bandwidth capabilities.

[0033] In implementation, the pre-scheduling and slice encapsulation modules can comprehensively consider the priority attributes of each VOQ cumulative buffer queue (such as latency-sensitive services, non-latency-sensitive services, elephant streams, mouse streams, etc.), the current cumulative number of packets in the VOQ cumulative buffer queue, and the status of the downstream egress queue level to perform differentiated dequeue scheduling. For example, dequeue triggering conditions can be set according to service priority, with first-priority services triggering dequeueing and second-priority services waiting to accumulate packets to a set threshold before being dequeueed in batches; first-priority services are higher than second-priority services; the timing of packet dequeueing can be adjusted in conjunction with the downstream queue level, and the dequeue rate can be adapted according to the downstream load status; then, the dequeueed service data is encapsulated into fixed-length cells to facilitate subsequent optoelectronic hybrid scheduling and switching processing.

[0034] Specifically, processing is differentiated according to the priority attributes of the services. For high-priority, latency-sensitive, and low-volume services, dequeueing can be performed directly without waiting for packet accumulation. The dwell time in the VOQ accumulation buffer queue is relatively short, and dequeueing is quick. The corresponding payload packet in each cell is less filled, resulting in lower latency. For low-priority, high-volume services (such as Elephant Stream), dequeueing is centralized after the packets accumulate to a preset packet accumulation threshold. The payload in each cell is more fully filled, resulting in higher encapsulation efficiency. In addition, the dequeueing timing can be dynamically adjusted according to the downstream egress queue level. When the downstream egress queue level is lower than the first set value (i.e., low) and bandwidth resources are sufficient, dequeueing and encapsulation are performed as soon as a packet arrives, resulting in lower latency, but also lower cell encapsulation efficiency. When the downstream egress queue level is higher than the second set value (i.e., high) and bandwidth resources are bottlenecked, dequeueing and encapsulation are performed uniformly after the packets accumulate to the preset packet accumulation threshold. Cell encapsulation efficiency is higher, but latency is relatively higher. In addition, priority attribute scheduling can be combined with downstream queue level scheduling to complete the pre-scheduling and dequeueing of the cumulative cache queue using a hybrid scheduling mode.

[0035] In this invention, a cell is a fixed-length packet of a predetermined size, carrying a cell header (HD) containing a sequence number, service traffic type, forwarding table, and payload data. A cell may contain one or more data packets. The advantage of cell encapsulation is that for services requiring electrical packet switching, normalized lengths allow for more efficient downstream electrical packet switching, increasing system switching capacity. For services requiring optical switching, fixed-length cells enable more precise time-gate-based control. Furthermore, this invention unifies optical and electrical switching services into cell slices and encapsulates them with sequence numbers, allowing for flexible load balancing between the same or different flows on the electrical packet switching unit (EPS) and the optical switching unit (OCS). This supports out-of-order recovery in hybrid optoelectronic switching, significantly expanding system switching capacity, providing high flexibility, and reducing system power consumption. The encapsulated cell enters the corresponding output buffer queue of the VOM output buffer module; this circuit is placed closest to the physical port on the system side to minimize delay jitter, so as to accurately control the time window of the output cell; each VOM output buffer queue contains at least one electrical path queue or optical path queue pointing to the destination unit, or both.

[0036] The aforementioned preset traffic allocation rules select whether to send encapsulated fixed-length cells to the electrical path queue or the optical path queue based on traffic characteristics and queue load status. Specifically, these rules can include the following: one is traffic allocation by flow type, such as assigning small-volume "mouse" flows to the electrical path queue and large-volume "elephant" flows to the optical path queue; another is preferential allocation within the same path, where the same service flow is preferentially assigned to the optical path queue; if the optical path queue is nearing full, it is switched to the electrical path queue, and vice versa; the third is dynamic traffic allocation based on bandwidth authorization, where the path assignment is not pre-fixed, and the decision of whether a cell enters the electrical path queue or the optical path queue is dynamically made through a bandwidth authorization mechanism.

[0037] The VOM egress buffer queue maintains the electrical path queue or optical path queue for each virtual outgoing unit. The optoelectronic hybrid scheduling module schedules the queues according to certain scheduling strategies based on information such as the VOM egress buffer queue type, queue level (i.e., water level information), optoelectronic bandwidth gating list, and backpressure at the outgoing port, outputting the data to the physical port. When a VOM queue is nearly full, backpressure is applied to the upstream pre-scheduling and slicing encapsulation modules, and the upstream VOQ cumulative buffer stops dequeuing. The electrical bandwidth gating list has statistical multiplexing characteristics, meaning multiple forwarding paths can be established within the same time slot; therefore, electrical scheduling can handle multiple VOMs within a single time slot. The optical bandwidth gating list has time-division multiplexing characteristics, meaning at most one optical forwarding path can be established within the same time slot; therefore, optical scheduling processes only a single VOM within a single time slot.

[0038] By using the aforementioned cell encapsulation and serial number introduction method, this invention can flexibly distribute the load of the same service across the optical switching unit (OCS) and the electrical packet switching unit (EPS), greatly expanding the system's switching capacity, and with lower power consumption compared to a standalone EPS.

[0039] Furthermore, in specific implementation, in the traffic control method provided in the embodiments of the present invention, each virtual outgoing unit (VOM) contains multiple virtual physical ports, and each virtual physical port can have multiple priorities, i.e., corresponding to multiple VOQ cumulative cache queues; while performing pre-scheduling and dequeuing according to the differentiated scheduling strategy in step S102, it may also include: scheduling in a time-division round-robin manner among the virtual physical ports, and scheduling in a priority manner among the priority queues under the same virtual physical port; performing multi-level round-robin scheduling at a set scheduling frequency, merging and converging the outgoing cache queues, and after convergence, performing queuing management only based on the outgoing service processing unit dimension in the outgoing cache queues.

[0040] During implementation, while performing pre-scheduled dequeueing, convergence from the VOQ accumulator queue to the VOM egress queue can also be executed. The VOM in the inbound service processing unit (IM) is a virtual logical unit configured with virtual physical ports; the outbound service processing unit (OM) carries real physical ports. For example, a 400G outbound service processing unit (OM) can be configured as 1×400G or 4×100G. Each physical port can contain 8 priority queues. The number of VOQ accumulator queues = number of outbound service processing units × number of physical ports per outbound service processing unit × number of priority queues under a single physical port, resulting in a relatively large number of queues.

[0041] To reduce timing pressure, pre-scheduling can perform multi-level polling at a set frequency (such as a frequency below a preset threshold): TDM time-division polling is performed between virtual physical ports in each virtual outgoing unit (VOM); SP polling is performed between priority queues under the same virtual physical port; and RR polling is used between service units. After multi-level polling scheduling, the number of queues is significantly reduced. After convergence, the granularity of virtual physical ports and priorities is no longer refined; unified queuing management is performed only at the outgoing service processing unit level in the VOM egress buffer queue, significantly reducing the system's timing processing pressure. At the same time, through packet accumulation strategy, large-capacity bandwidth processing can be performed at a lower scheduling frequency, reducing circuit timing pressure, greatly improving system processing bandwidth, and balancing the low latency of latency-sensitive services and the high bandwidth efficiency of elephant streams during the scheduling process.

[0042] Furthermore, in specific implementation, in the traffic control method provided in the embodiments of the present invention, step S103 maintains the optical and electrical bandwidth gating list and initiates a bandwidth request to the destination processing end when the remaining bandwidth does not meet the service carrying requirements. Specifically, this may include: using a time-aware shaping mechanism (TAS) to maintain the bandwidth of the optical bandwidth gating list and using a token-based shaping mechanism to maintain the bandwidth of the electrical bandwidth gating list, wherein the number of token bits that can be obtained by the electrical bandwidth gating list is calculated and updated based on the authorized time slot and port rate; detecting the remaining bandwidth resources of the optical bandwidth gating list and the electrical bandwidth gating list; when the remaining bandwidth resources do not meet the service carrying requirements, initiating a bandwidth resource request to the destination processing end so that the destination processing end issues bandwidth authorization based on its own egress cache load status, and stopping the issuance of bandwidth authorization when the egress cache reaches a preset load threshold.

[0043] like Figure 3As shown, the Incoming Service Processing Unit (IM) can also include a multi-mode bandwidth request module. This module maintains an optical and electrical bandwidth gating list and requests bandwidth from the destination processing unit when insufficient remaining bandwidth is detected. The optical bandwidth gating list can be maintained based on a mechanism similar to TAS, while the electrical bandwidth gating list can be maintained based on a token mechanism similar to CBS (Credit-Based Shaper). The relationship between tokens and time slots can be obtained using the following formula: Number of token bits obtained by the bandwidth gating list = Authorized time slot × Port rate.

[0044] Furthermore, in specific implementation, in the above-mentioned traffic control method provided in the embodiments of the present invention, after executing step S103 to initiate a bandwidth request to the destination processing end, it may further include: the destination processing end periodically generates bandwidth authorization resources according to the system port rate, and uses the product of the time slot and the port rate as the total number of bits authorized in a single instance; if the bandwidth request is an electrical switching path bandwidth request, then a statistical multiplexing method is used for resource allocation, and the current time slot resources are evenly divided or allocated to each electrical switching path bandwidth request according to the queue level weight; if the bandwidth request is an optical switching path bandwidth request, then a time-division multiplexing method is used for resource allocation, and only one optical switching path bandwidth request is authorized in each time slot, while the remaining optical switching path bandwidth requests are converted into electrical switching path bandwidth requests to participate in statistical multiplexing, or postponed to the next time slot to participate in authorization competition again; if there are electrical switching path bandwidth requests and optical switching path bandwidth requests in the same time slot, then one is authorized according to a preset priority rule, or the bandwidth authorization allocation of electrical switching path bandwidth requests and optical switching path bandwidth requests is completed in time-division interleaving.

[0045] In implementation, the outbound service processing unit (OM) may include a multi-mode bandwidth authorization module. This module periodically generates bandwidth gating based on the system port rate and a certain scale; the number of bits authorized is represented by a timeslot multiplied by the port rate. It also receives bandwidth request requests from all incoming electrical and optical switching paths, and issues bandwidth authorizations based on its own egress buffer load status. Bandwidth authorization issuance stops when the egress buffer reaches a preset load threshold. Within the multi-mode bandwidth authorization module, electrical switching path requests are decided using a statistical multiplexing method, which can evenly distribute the current timeslot or allocate it to all electrical requests according to queue weights. Optical switching path requests are decided using a time-division multiplexing method, with each timeslot at most authorized to one optical request; the remaining requests are converted to electrical authorizations for statistical multiplexing or wait for re-authorization in the next timeslot. If both electrical and optical requests exist simultaneously, one can be prioritized for authorization according to specific rules, or a time-division interleaving method can be used for authorization.

[0046] Furthermore, in specific implementation, in the above-mentioned traffic control method provided in the embodiments of the present invention, after receiving the bandwidth authorization from the destination processing end, step S104 updates the bandwidth resources and controls the cells in the egress buffer queue to be forwarded through the electrical packet switching unit according to the electrical gating state, and controls the cells in the egress buffer queue to be forwarded through the optical switching unit according to the optical gating state. Specifically, it may include: after receiving the bandwidth authorization instruction issued by the destination processing end, updating the remaining bandwidth resources in the optoelectronic bandwidth gating list, and supplementing the authorized bandwidth quota to the corresponding time slot or token number list; querying the gating state of the optoelectronic bandwidth gating list; if the optical gating state is open, controlling the cells in the egress buffer queue to be sent into the optical switching unit, and the optical switching unit to complete the forwarding in a time-division exclusive manner; if the electrical gating state is open, controlling the cells in the egress buffer queue to be sent into the electrical packet switching unit, and the electrical packet switching unit to complete the forwarding in a statistical multiplexing manner; if both the optical gating state and the electrical gating state are closed, then prohibiting the cells in the egress buffer queue from dequeuing, suspending the forwarding operation of the relevant cells, until the gating state switches to open.

[0047] In implementation, after receiving the bandwidth authorization instruction, the Incoming Service Processing Unit (IM) first updates the remaining bandwidth resources in the optical and electrical bandwidth gating list, supplementing the authorized bandwidth quota to the corresponding time slot or token count list; then it queries the gating status of the optical and electrical bandwidth gating list. If the optical gating status is open, it controls the cells in the egress buffer queue to be sent to the Optical Switching Unit (OCS), which then forwards them in a time-division exclusive manner; if the electrical gating status is open, it controls the cells in the egress buffer queue to be sent to the Electrical Packet Switching Unit (EPS), which then forwards them in a statistical multiplexing manner; if both the optical and electrical gating statuses are closed, it prohibits the dequeueing of cells in the corresponding queue and suspends the forwarding operation of the relevant cells until the gating status switches to open.

[0048] Under this mechanism, for each cell dequeued, the electrical gating list can deduct the number of bits corresponding to the cell frame length. The electrical gating list can obtain multiple VOM electrical grants within a time slot, while the optical gating list can only obtain one VOM optical grant within a time slot. Based on this rule, each time slot establishes at most one unique optical path and multiple statistically multiplexed electrical paths between the inbound service processing unit IM_i and the outbound service processing unit OM_j, allowing for flexible coordination of optical and electrical bandwidth.

[0049] To achieve the selection of a unique optical path, two implementation schemes can be adopted: One scheme involves determining the optical path when the IM (Initial Message Block) initiates the request. If multiple VOM (Vendor Message Block) optical requests exist, the one with the highest queue level is selected as the optical request, and the others are electrical requests. At the outgoing Service Processing Unit (OM), the request messages from different incoming Service Processing Units (IMs) are selected, with the one with the highest queue level receiving optical authorization, thus obtaining a unique optical path. The other requests are statistically multiplexed as electrical paths. The other scheme does not perform the determination at the time of request initiation, but carries queue level information in the request message. The outgoing Service Processing Unit (OM) selects from multiple incoming Service Processing Unit (IM) request messages, and the incoming Service Processing Unit (IM) selects from multiple destination outgoing Service Processing Unit (OM) authorization messages, obtaining a unique optical path. The others are multiplexed as electrical paths.

[0050] The electrical packet switching unit (EPS) is an n×n packet switching matrix that performs packet or cell-based packet switching on all incoming traffic to the outgoing traffic, and handles the forwarding of all optoelectronic bandwidth requests and bandwidth grant messages. Figure 4 This is a bandwidth request and authorization path diagram provided for embodiments of the present invention. For example... Figure 4 As shown, the present invention can perform necessary time synchronization for each unit, maintain the optical gating window of each VOM time slot by obtaining bandwidth request authorization messages, and notify the optical switching unit OCS to perform fast optical path switching.

[0051] The Optical Switching Unit (OCS) primarily provides multiple optical input / output service interfaces and at least one control interface. The control interface receives external configuration commands and configures the connection relationships between the input and output of the optical service interfaces. Once configured, optical services input from the optical service interfaces are forwarded to the corresponding optical output interfaces according to the configuration. This can be achieved using methods such as Micro-Electro-Mechanical Systems (MEMS), Planar Lightwave Circuits (PLCs), and Wavelength Selective Switches (WSS). The control channel can be quickly configured via an Electrical Packet Switching Unit (EPS) or other control planes.

[0052] This invention can flexibly allocate and dynamically adjust the flow of optical switching paths and electrical switching paths according to the different traffic characteristics of services, making full use of the statistical multiplexing characteristics of EPS electrical packet services and the time-division multiplexing characteristics of OCS optical switching services, thereby effectively improving the overall bandwidth utilization.

[0053] Furthermore, in specific implementation, in the above-mentioned traffic control method provided in the embodiments of the present invention, step S105 decapsulates and reassembles the data stream of the information cell, and outputs it after scheduling by the service-side egress buffer. Specifically, it may include: decapsulating the information cell, stripping the control and sequence information of the information cell header, and extracting the service payload data carried inside the information cell; using the sequence number carried in the information cell header to sort and organize the unordered payload data to restore the original service data stream; sending the restored original service data stream to the service-side egress buffer for temporary storage, and performing dequeue scheduling on the data in the service-side egress buffer according to the set scheduling strategy to complete the output of service data; monitoring the load of the egress buffer, and when the load of the egress buffer reaches a preset load threshold, sending a backpressure signal upstream until the buffer load falls back to the preset range.

[0054] It should be noted that the outbound service processing unit (OM) may also include an optoelectronic hybrid packet reassembly module, an egress buffer module, and an egress scheduling module. The optoelectronic hybrid packet reassembly module can uniformly decapsulate cells from both electrical and optical switching interfaces, stripping the control fields and sequence number information from the cell headers to extract the service payload data carried by the cells. Based on the sequence numbers carried by the cells, it sorts and organizes the out-of-order service payloads, restoring the complete original service data stream, and writes the reassembled service data stream into the egress buffer module. The egress scheduling module, according to a preset scheduling strategy, dequeues the data in the egress buffer module in an orderly manner, completing the outbound output of the service data.

[0055] Figure 5 This is a block diagram illustrating the principle of optoelectronic hybrid bandwidth application and authorization provided in an embodiment of the present invention. Figure 5 As shown, this invention monitors the occupancy status of the egress cache queue in real time. When the egress cache queue approaches full capacity and the load reaches a preset threshold, backpressure is applied to the multi-mode bandwidth authorization module. At this time, bandwidth allocation is not allowed, and bandwidth authorization is stopped until the load on the egress cache module decreases and the backpressure is removed, at which point the normal bandwidth authorization process resumes. The bandwidth allocation rate of the system port needs to have a certain speedup compared to the user port, mainly considering the efficiency loss of cell encapsulation and the jitter of upstream and downstream interactions, in order to achieve line-speed processing.

[0056] Figure 6 This is a flowchart illustrating the photoelectric hybrid flow control provided in an embodiment of the present invention. Figure 6As shown, data services first arrive at the Inbound Service Processing Unit (IM), and after flow parsing and classification, enter the VOQ (Voice over Query) accumulation buffer queue. When the queue data volume accumulates to a threshold, the system pre-schedules the data out of the queue according to a differentiated strategy, encapsulates the data into fixed-length cells, and distributes them to the optical path queue or electrical path queue in the VOM (Voice over Query) egress buffer queue. The multi-mode bandwidth application module maintains the time slot list and manages the outbound window. When bandwidth is insufficient, it initiates a bandwidth application to the destination processing unit (OM). After authorization, the Inbound Service Processing Unit (IM) updates the time slot list. Subsequently, the optoelectronic hybrid scheduling module combines queue and time slot information, using statistical multiplexing scheduling for the electrical path queue and time-division multiplexing scheduling for the optical path queue. Telecommunication cells are forwarded to the destination processing unit (OM) via the EPS (Electrical Switching Unit), and optical cells are forwarded to the destination processing unit (OM) via the OCS (Optical Switching Unit). Finally, the destination processing unit (OM) performs unified decapsulation and data reassembly on the arriving optical and electrical path cells, and outputs complete service data after egress buffer scheduling. The entire process, through VOQ cumulative caching, differentiated scheduling, dynamic bandwidth application and authorization, optoelectronic path diversion and forwarding, and egress reassembly, achieves orderly management of service flows and coordinated allocation of optoelectronic resources, ensuring efficient and reliable transmission of different services.

[0057] The following uses a latency-sensitive service as an example to illustrate the traffic control method provided by this invention: For high-priority services with low latency sensitivity and small traffic, in order to minimize latency, multiple packets are not accumulated during VOQ accumulation. Instead, packets are encapsulated into cells as soon as they are available. Figure 7 This invention provides encapsulation strategies for latency-sensitive services in embodiments of the invention. For example... Figure 7 As shown, P0, P1, and Pn are quickly dequeued after entering the VOQ buffer. The first cell only carries data packet P0, the second cell only carries data packet P1, and the nth cell only carries data packet Pn. The cell header HD carries sequence numbers 0, 1, and n, as well as the delay-sensitive mouse stream type, and enters the electrical forwarding exit VOM queue to fully utilize the packet-based switching characteristics of the electrical packet switching unit EPS.

[0058] The electrical path queue scheduling module can decide on dequeueing multiple queues within each time slot, following statistical multiplexing characteristics. It can poll according to a round-robin (RR) strategy, performing cell dequeueing when the queue is not empty and bandwidth tokens allow. Simultaneously, if the remaining tokens fall below a certain threshold, the bandwidth request module initiates an electrical request. In the outbound service processing unit (OM) multi-mode bandwidth authorization module, the current bandwidth is gated and evenly authorized to all electrical request requests. When the VOM receives authorization, the number of token bits obtained is represented by multiplying the time value in the authorization message by the port rate. If the token is positive, dequeueing is allowed; after dequeueing, the corresponding token is deducted.

[0059] Figure 8This is a timing diagram illustrating the latency-sensitive service bandwidth authorization mechanism provided in an embodiment of the present invention. Figure 8 As shown, IM1, IM2, and IM3 each have a latency-sensitive mouse stream sent to the outbound service processing unit OM1, which enters the egress queue VOM1 and initiates an electrical bandwidth request. During S0, OM1 receives a total of three incoming electrical bandwidth request requests from IM1, IM2, and IM3. The multi-mode bandwidth authorization module authorizes 1 / 3 of the bandwidth for each IM. After receiving the bandwidth authorization, the bandwidth request module updates the remaining bandwidth list and allows dequeueing from VOM. IM1, IM2, and IM3 can simultaneously send packets to the electrical packet switching unit EPS to OM1, demonstrating the statistical multiplexing characteristics.

[0060] The following uses a high-bandwidth service as an example to illustrate the traffic control method provided by this invention: When the Elephant Stream service accumulates data for dequeueing in VOQ, the decision can be made in conjunction with the downstream VOM waterline. If the VOM waterline is relatively low, it means that the system-side bandwidth is sufficient, and data packets are dequeued and encapsulated as soon as they are available. At this time, the latency is low, but the cell encapsulation efficiency is also low. If the VOM waterline is relatively high, it means that the system-side bandwidth is relatively tight. Dequeueing is scheduled only after the data accumulates to a certain threshold or the waiting time expires, so as to fully fill the data and improve the cell encapsulation efficiency and save bandwidth. Since the Elephant Stream itself has a large traffic volume, the accumulated waiting latency is relatively small.

[0061] Figure 9 This is a schematic diagram illustrating the encapsulation strategy for the Elephant Stream service provided in an embodiment of the present invention. For example... Figure 9 As shown, data packets P0, P1, and Pn accumulate in the VOQ. Once they reach a threshold, they are encapsulated into the same cell, thus significantly reducing the pre-scheduling frequency. The cell header HD carries sequence number 0, indicating an elephant stream type, and prioritizes entry into the VOM queue.

[0062] The optical path queue scheduling module can only decide to dequeue one queue at most in each time slot, following the time-division multiplexing characteristics. It can schedule dequeues based on information such as the non-empty state of each VOM and whether the current time slot is open. At the same time, the bandwidth request module initiates an optical request for the next scheduling time slot. The optical request cell can carry FIFO watermark information to provide a basis for the decision of the outgoing service processing unit OM. When the VOM is authorized, it maintains the time slot gating list based on the time slot value inside.

[0063] Figure 10 This is a schematic diagram illustrating the bandwidth authorization mechanism decision-making for the elephant stream service provided in an embodiment of the present invention. Figure 10As shown, firstly, IM1 contains three optical bandwidth requests destined for OM1, OM2, and OM3, and IM2 contains one optical bandwidth request destined for OM1. Each carries queue watermark information or backlog cell count information. Specifically, IM1 has VOM_1 watermark level 2, VOM_2 watermark level 1, and VOM_3 watermark level 2, while IM2 has VOM_1 watermark level 3. Then, on OM1, two optical requests from IM1 and IM2 are received in time slot S0. IM2 has a higher watermark, so IM2 is selected for optical granting, and IM1 for electrical granting. OM2 and OM3 each have one optical request, both of which are optically granted. Afterward, IM1 receives the electrical grant from OM1, as well as the optical grants from OM2 and OM3. Since VOM3 has a higher watermark than VOM2, IM1 selects VOM3 to update the optical gating list and schedules the optical path to OM3 within day 1 (the time slot when the optical path is stable and available).

[0064] Figure 11 A timing diagram of the bandwidth authorization mechanism for the elephant stream service provided in an embodiment of the present invention. (See diagram below.) Figure 11 As shown, both VOM1 and VOM2 are updated to the electrical gating list and scheduled to the electrical path; IM2 updates the optical gating list of VOM1 and is scheduled to the optical path at day 1, switching to OM1. At the same time, when bandwidth request authorization messages pass through the electrical packet switching unit EPS, optical switching unit OCS configuration information is formed on the electrical packet switching unit EPS and quickly sent to the optical switching unit OCS for optical line reconfiguration.

[0065] The above example schedules services with relatively larger bandwidth to the Optical Switching Unit (OCS) for switching, while other services are scheduled to the Electrical Packet Switching Unit (EPS). By making full use of the statistical multiplexing characteristics of EPS electrical packet services and the time-division multiplexing characteristics of OCS optical switching services, service congestion can be effectively reduced and bandwidth utilization can be improved.

[0066] The following uses a mixed service as an example to illustrate the traffic control method provided by this invention: In a real network, there are various types of flows, such as high-frequency, low-volume latency-sensitive control flows, and low-frequency, high-bandwidth AI training aggregation communication flows, which may go to multiple destination units. After flow parsing and classification, the flows enter different priority VOQ queues. The flow is processed comprehensively based on the service priority and the waterline of the downstream egress VOM queue. When the egress VOM queue waterline is low or the service is high priority, no accumulation is required; cells are encapsulated as soon as packets are available. When the egress VOM queue waterline is high or the service is low priority, the flow is allowed to accumulate to a certain threshold, or the waiting time expires before being scheduled for dequeueing and cell encapsulation. When the egress VOM queue is nearly full, VOQ dequeueing is not allowed; low-volume cells are given priority to enter the electrical egress VOM queue; high-volume cells are given priority to enter the optical egress VOM queue, and when the optical egress VOM queue is nearly full, they enter the electrical path queue. When the VOM optical path queue or electrical path queue is not empty and the remaining bandwidth is insufficient, an optical or electrical bandwidth request is initiated. The destination processing end collects all electrical requests and evenly allocates the current time slot to all electrical request requests; it also collects all optical requests and, according to certain rules, such as using the watermark in the request message as a priority decision, allocates the current time slot to a single optical request request; if both optical and electrical requests exist in the current time slot, the selection is made according to the watermark priority, or time-division multiplexing authorization is performed, that is, optical authorization is performed during the day (available time slots) of optical bandwidth, and electrical authorization is performed during the night (unavailable time slots, including system reconfiguration time).

[0067] Figure 12 This is a timing diagram of the hybrid mode bandwidth licensing mechanism provided in an embodiment of the present invention. Figure 12As shown, there is an elephant stream on IM1 destined for OM1, and a mouse stream on IM2 destined for OM1. Service flow cells on IM1 first enter the VOM1 optical path queue. When the optical path queue reaches a certain threshold, they enter the electrical path queue. Therefore, optical requests and electrical requests are sent to OM1 respectively. Mouse stream cells on IM2 directly enter the VOM1 electrical path queue and initiate electrical requests. The multi-mode bandwidth granting module on OM1 periodically generates bandwidth time slots Si. Optical requests from IM1 and electrical requests from IM1 and IM2 are received in time slot S0. The day bandwidth of S0 is granted to the optical request of IM1, and it is instructed to send packets in the day1 time window. The night bandwidth is granted evenly or according to the waterline weight to the electrical requests of IM1 and IM2. After IM1 receives the authorization packet, the bandwidth request module updates the remaining bandwidth list. The optical path queue gating allows the optical path queue to be dequeued from VOM1 during the day1 time period. The electrical path queue receives an update of the token bit count obtained by multiplying the night time slot by the port rate and immediately allows VOM1 electrical path queue scheduling to be dequeued. IM2 only updates the electrical path queue token and immediately allows VOM1 electrical path queue scheduling to be dequeued. This example, through dynamic control of optical and electrical traffic, fully utilizes the time during which the optical switching unit (OCS) reconfigures to guide services into the electrical packet switching unit (EPS). This allows a service flow to be load-shared on both the OCS and EPS, improving bandwidth utilization. At the same time, latency-sensitive mouse flows can be quickly forwarded through the EPS without being blocked by elephant flows, reducing latency.

[0068] From the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by software plus necessary general-purpose hardware platforms, and of course, they can also be implemented by hardware, but in many cases the former is a better implementation method.

[0069] Embodiments of the present invention also provide a flow control device. Based on the functional module perspective, this device is applied to a hybrid optoelectronic switching system and includes: The incoming service processing unit is used to perform flow parsing and classification on the incoming data services and store them in the cumulative buffer queue according to the service flow type. When the amount of data in the cumulative buffer queue reaches the preset message accumulation threshold, it performs pre-scheduling and dequeueing according to the differentiated scheduling strategy, encapsulates the dequeueed data into fixed-length cells, and stores them in the outgoing buffer queue. It maintains the optoelectronic bandwidth gating list and initiates a bandwidth application to the destination processing end when the remaining bandwidth does not meet the service carrying requirements. After receiving the bandwidth authorization issued by the destination processing end, it updates the bandwidth resources and controls the cells in the outgoing buffer queue to be forwarded through the electrical packet switching unit according to the electrical gating status, and controls the cells in the outgoing buffer queue to be forwarded through the optical switching unit according to the optical gating status. The electrical packet switching unit is used to receive cells sent to the incoming service processing unit and forward the cells using statistical multiplexing. The optical switching unit is used to receive cells sent by the incoming service processing unit and forward the cells in a time-division exclusive mode. The outbound service processing unit is used to receive cells forwarded by the electrical packet switching unit and the optical switching unit, decapsulate and reassemble the data streams of the cells, and output them outward after being scheduled by the service-side egress buffer; it is also used to receive bandwidth requests from the inbound service processing unit, issue bandwidth authorizations so that the inbound service processing unit can update bandwidth resources and manage cell forwarding according to the gating status; the outbound service processing unit contains multiple processing ends, and the destination processing end is any one of the processing ends in the outbound service processing unit.

[0070] In the traffic control device provided in this embodiment of the invention, access data services can be classified and stored in an accumulated cache queue according to the service flow type. When the data volume in the accumulated cache queue reaches a preset message accumulation threshold, pre-scheduling and dequeuing are performed according to a differentiated scheduling strategy. The dequeued data is encapsulated into fixed-length cells and stored in the egress cache queue. By maintaining an optoelectronic bandwidth gating list, when bandwidth resources are insufficient, a bandwidth request is initiated to the destination processing end. After receiving bandwidth authorization, the bandwidth resources are updated, and the cells in the egress cache queue are forwarded via the electrical packet switching unit according to the electrical gating state, and the cells in the egress cache queue are forwarded via the optical switching unit according to the optical gating state. This allows for dynamic and flexible allocation of optoelectronic switching bandwidth resources and real-time dynamic adjustment, significantly improving the overall bandwidth utilization. In addition, optoelectronic switching services are uniformly encapsulated by cell slicing, supporting flexible load sharing between the same service flow or different service flows on the electrical packet switching and optical switching paths, effectively saving system resources and operating power consumption. After the cells are forwarded to the destination processing end, unified decapsulation, data flow reassembly, and egress cache scheduling output further ensure the stability of service transmission.

[0071] Since the embodiments of the flow control device and the flow control method correspond to each other, the descriptions of the features in the embodiment corresponding to the flow control device can be found in the relevant descriptions of the embodiment corresponding to the flow control method, and will not be repeated here. Furthermore, it has the same beneficial effects as the flow control method mentioned above.

[0072] Embodiments of the present invention also provide an electronic device, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the above-described traffic control method embodiments.

[0073] Embodiments of the present invention also provide a computer-readable storage medium storing a computer program configured to execute the steps in any of the above-described traffic control method embodiments when running.

[0074] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0075] Embodiments of the present invention also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the above-described traffic control method embodiments.

[0076] Embodiments of the present invention also provide another computer program product, including a non-volatile computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps in any of the above-described traffic control method embodiments.

[0077] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0078] The present invention has provided a detailed description of a flow control method, apparatus, device, and medium. Specific examples have been used to illustrate the principles and implementation methods of the invention. The descriptions of these embodiments are only intended to aid in understanding the method and core ideas of the present invention. It should be noted that those skilled in the art can make various improvements and modifications to the present invention without departing from its principles, and these improvements and modifications also fall within the protection scope of the present invention.

Claims

1. A traffic control method, characterized in that, Applications include: (The text abruptly ends here, so the translation also ends here.) The incoming data services are stream parsed and classified, and stored in the cumulative cache queue according to the service flow type; When the amount of data in the cumulative buffer queue reaches the preset message accumulation threshold, pre-schedule dequeueing is performed according to the differentiated scheduling strategy, and the dequeueed data is encapsulated into fixed-length information cells and stored in the exit buffer queue. Maintain the optoelectronic bandwidth gating list and initiate a bandwidth request to the destination processing terminal when the remaining bandwidth does not meet the service carrying requirements; After receiving the bandwidth authorization from the destination processing end, the bandwidth resources are updated, and the cells of the egress buffer queue are forwarded via the electrical packet switching unit according to the electrical gating state, and the cells of the egress buffer queue are forwarded via the optical switching unit according to the optical gating state. After the information cell is forwarded to the destination processing end, the information cell is decapsulated and the data stream is reassembled, and then output after being scheduled by the service-side egress cache.

2. The traffic control method according to claim 1, characterized in that, Pre-scheduled dequeueing is performed according to a differentiated scheduling strategy. The dequeueed data is encapsulated into fixed-length cells and stored in the egress buffer queue, including: Based on the priority of the accumulated cache queue, the accumulated number of messages in the queue, and the downstream exit queue level information, a differentiated scheduling strategy is adopted to perform pre-scheduled dequeueing from the accumulated cache queue; The outgoing business data is encapsulated and processed to generate fixed-length information cells carrying sequence numbers, business flow types, and forwarding tables; The fixed-length cells are stored in the electrical path queue or optical path queue in the egress buffer queue according to a preset diversion rule that includes traffic size attributes, queue load status, or path selection rules.

3. The traffic control method according to claim 1, characterized in that, Maintain the optoelectronic bandwidth gating list and initiate bandwidth requests to the destination processing terminal when the remaining bandwidth does not meet the service carrying requirements, including: A time-aware shaping mechanism is used to maintain the bandwidth of the optical bandwidth gating list, and a token-based shaping mechanism is used to maintain the bandwidth of the electrical bandwidth gating list. The number of token bits that can be obtained in the electrical bandwidth gating list is calculated and updated based on the authorized time slot and port rate. Detect the remaining bandwidth resources in the optical bandwidth gating list and the electrical bandwidth gating list; When the remaining bandwidth resources do not meet the service carrying requirements, a bandwidth resource request is initiated to the destination processing terminal, so that the destination processing terminal issues bandwidth authorization based on its own egress cache load status. When the egress cache reaches the preset load threshold, the issuance of bandwidth authorization is stopped.

4. The traffic control method according to claim 1, characterized in that, After initiating a bandwidth request to the destination processor, the process also includes: The target processing terminal periodically generates bandwidth authorization resources based on the system port rate, and uses the product of the time slot and the port rate as the total number of bits authorized in a single transaction. If the bandwidth request is for an electrical switching path, then the resource allocation is carried out using a statistical multiplexing method, which evenly divides the current time slot resources or allocates them to each electrical switching path bandwidth request according to the queue level weight. If the bandwidth request is for an optical switching path, then time-division multiplexing is used for resource allocation. Only one optical switching path bandwidth request is authorized per time slot, and the remaining optical switching path bandwidth requests are converted into electrical switching path bandwidth requests to participate in statistical multiplexing, or are postponed to the next time slot to participate in the authorization competition again. If both electrical switching path bandwidth requests and optical switching path bandwidth requests exist simultaneously in the same time slot, one will be authorized according to the preset priority rules, or the bandwidth authorization allocation for electrical switching path bandwidth requests and optical switching path bandwidth requests will be completed in time-division interleaving mode.

5. The traffic control method according to claim 1, characterized in that, After receiving the bandwidth authorization from the destination processing end, the bandwidth resources are updated, and the cells in the egress buffer queue are forwarded via the electrical packet switching unit according to the electrical gating state, and the cells in the egress buffer queue are forwarded via the optical switching unit according to the optical gating state, including: After receiving the bandwidth authorization instruction issued by the target processing terminal, update the remaining bandwidth resources in the optoelectronic bandwidth gating list and supplement the authorized bandwidth quota to the corresponding time slot or token number list. Query the gating status of the optoelectronic bandwidth gating list; If the optical gating state is open, the cells in the egress buffer queue are sent to the optical switching unit, which then forwards them in a time-division exclusive manner. If the electrical gating state is open, the cells in the egress buffer queue are sent to the electrical packet switching unit, which then forwards them in a statistical multiplexing manner. If both the optical gating state and the electrical gating state are closed, then cell dequeueing from the exit buffer queue is prohibited, and the forwarding operation of the relevant cells is suspended until the gating state switches to open.

6. The traffic control method according to claim 1, characterized in that, The information cells are decapsulated and the data stream is reassembled, and then output after being scheduled by the service-side egress buffer, including: The cell is decapsulated to remove the control and sequence information from the cell header and extract the service payload data carried inside the cell. By using the sequence number carried in the cell header, the unordered payload data is sorted and organized to restore the original business data stream; The restored original business data stream is sent to the business-side egress cache for temporary storage. The data in the business-side egress cache is dequeued and scheduled according to the set scheduling strategy to complete the output of business data. Monitor the load of the egress cache. When the load of the egress cache reaches the preset load threshold, send a backpressure signal upstream to adjust the multi-mode bandwidth authorization rate until the cache load falls back to the preset range.

7. The traffic control method according to claim 1, characterized in that, The incoming data services are stream parsed and classified, and stored in an accumulated cache queue according to the service flow type, including: Receive data services from external access and read message-related fields and traffic information from the data services; Perform flow parsing and service classification on the read message-related fields and traffic information to identify service flow types with different priorities; The data services are stored sequentially into the corresponding priority cumulative cache queue according to the priority of the service flow type.

8. A flow control device, characterized in that, Applications include: (The text abruptly ends here, so the translation also ends here.) The incoming service processing unit is used to perform flow parsing and classification on the incoming data services and store them in the cumulative cache queue according to the service flow type. When the data volume of the cumulative cache queue reaches the preset message accumulation threshold, it performs pre-scheduling and dequeueing according to the differentiated scheduling strategy, encapsulates the dequeueed data into fixed-length cells, and stores them in the outgoing cache queue. It maintains the optoelectronic bandwidth gating list and initiates a bandwidth application to the destination processing end when the remaining bandwidth does not meet the service carrying requirements. After receiving the bandwidth authorization issued by the destination processing end, it updates the bandwidth resources and controls the cells in the outgoing cache queue to be forwarded through the electrical packet switching unit according to the electrical gating state, and controls the cells in the outgoing cache queue to be forwarded through the optical switching unit according to the optical gating state. The electrical packet switching unit is used to receive the cells sent by the incoming service processing unit and to forward the cells in a statistical multiplexing manner. The optical switching unit is used to receive the cells sent by the incoming service processing unit and forward the cells in a time-division exclusive mode. The outbound service processing unit is used to receive cells that arrive after being forwarded by the electrical packet switching unit and the optical switching unit, decapsulate and reassemble the data streams of the cells, and output them outward after being scheduled by the service-side egress buffer; it is also used to receive bandwidth requests from the inbound service processing unit and issue bandwidth authorizations; the outbound service processing unit includes multiple processing ends, and the destination processing end is any one of the processing ends in the outbound service processing unit.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the steps of the traffic control method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the traffic control method as described in any one of claims 1 to 7.

Images