A method and system for dynamic matching of circuit resources for different network exchanges

Abstract

The application discloses a circuit resource dynamic matching method and system for different network exchanges, and specifically digital information transmission, solves the mapping instability caused by the version dislocation of the gating or equivalent scheduling table between adjacent nodes in the update window during the configuration update, and is achieved by the following steps: a control plane distributes a window mark, collects node versions and adjacency abstracts, generates an alignment view, a data plane generates a shadow gating table, outputs a switching token according to a consistent switching judgment, triggers the upstream and downstream to simultaneously enable the shadow gating table and reserve the old table coverage by the token, returns an edge confirmation signal from the downstream in the first effective window after the enabling, compares the actual output with the shadow gating table, performs rollback and marking, finally collects the confirmation signal and the alignment view to form a stability record, solidifies the update sequence and the reservation rule, and converges the mapping instability in the transition period.

Description

TECHNICAL FIELD

[0001] The application relates to a circuit resource dynamic matching method and system for different network exchanges. BACKGROUND

[0002] In different network exchange scenarios, in order to obtain stable port, queue and time slot correspondence, key services often use gating or equivalent time-aware scheduling mechanisms to organize in-out queue windows on multi-type switching devices, aiming to maintain the continuity of the queuing and release order along the way. Offline generation and verification of the scheduling table are commonly used in engineering practice, and then the control plane distributes and enables the scheduling table at the reference time. With the expansion of network size and the increase of multi-domain interconnection, the differences in configuration propagation, driving load delay and node clock micro-offset are often superimposed, and the new and old tables coexist in the short window state of the adjacent nodes in the update window. Services are more likely to trigger edge misalignment on cross-domain or cross-layer paths, and then short-time interruption, slight out-of-order and jitter accumulation occur. The traditional method focuses on version consistency and timing activation, and lacks special verification of the continuity and edge fitting of actual out-of-order behavior, and it is difficult to maintain stable mapping of circuit resources under multi-domain conditions.

[0003] However, in different network exchange environments, the circuit resource dynamic matching is unstable due to the inconsistency of the scheduling table version and the edge phase of adjacent nodes in the update window. Typical triggering occurs in the process of batch and domain delivery and segmented activation: the adjacent hops in the path enter the running state at different times, and there is a short-time difference in the loading of table items and the port out-of-order scheduling. The micro-offset of the cross-domain clock causes the opening and closing edges to be disconnected in the transition window. The direct result is that the ternary correspondence of the port, queue and time slot is pulled apart, and the upstream released edge encounters a closed state or is released in advance at the downstream, forming truncation or stretching; when carrying control instructions, sampling data or low-jitter services, the interruption and out-of-order induced by misalignment will be amplified to the upper-layer application link, weakening the predictability and continuity of the network. In order to converge the above mismatch, it is necessary to build an alignment view in the control plane and deliver window markers, and in the data plane, to complete the controlled activation by using shadow gating tables to cooperate consistent switching judgment, and to close the verification chain by using edge confirmation signals and stability records, so as to smoothly return to the normal state from the transition period.

[0004] In order to solve the above problems, a technical scheme is provided. SUMMARY

[0005] In order to overcome the above-mentioned defects of the prior art, embodiments of the present application provide a circuit resource dynamic matching method and system for different network switching, which generates an alignment view by distributing window marks in the control plane and summarizing node gate table versions and adjacency abstracts, generates a shadow gate table in the data plane, and generates a switching token according to consistent switching determination to lock ports and queues, triggers the shadow gate table to be enabled simultaneously in the upstream and downstream by the switching token and retains the old gate table coverage, performs rollback and marking in the first valid window after enabling by comparing actual dequeuing in the upstream with the shadow gate table according to the edge confirmation signal returned by the downstream, generates a stability record by summarizing the edge confirmation signal and the alignment view, and solidifies the next batch of update order and the rule of frozen window mark to solve the mapping instability problem in the update window mentioned in the background art.

[0006] To achieve the above object, the present application provides the following technical scheme:

[0007] A circuit resource dynamic matching method for different network switching, comprising the steps of:

[0008] distributing window marks in the control plane and summarizing gate table versions and adjacency abstracts of each node to generate an alignment view as a prerequisite for consistent switching;

[0009] generating an inactivated shadow gate table in the data plane, selecting locked ports and queues according to consistent switching determination results, and generating a switching token, or performing delayed review, or directly rejecting triggering;

[0010] using the switching token as the only trigger to enable the shadow gate table simultaneously in the upstream and downstream, and retaining the old gate table coverage update window to maintain continuous circuit mapping and uninterrupted dequeuing;

[0011] in the first valid window after enabling, returning an edge confirmation signal by the downstream, comparing actual dequeuing in the upstream with the shadow gate table, and if a deviation is found, performing single rollback and marking an error jump point as a correction basis;

[0012] generating a stability record by summarizing the edge confirmation signal and the alignment view, solidifying the next batch of update order and the rule of frozen window mark according to the record, and freezing the window mark to complete the current update.

[0013] In a preferred embodiment, the gate table versions and adjacency abstracts of each node are distributed in the control plane and summarized to generate an alignment view as a prerequisite for consistent switching, wherein the control plane broadcasts update intentions and window marks to participating nodes, and the nodes read local gate table versions and version abstracts of adjacency ports and report them.

[0014] In a preferred embodiment, the control plane reports to generate an aligned view of the cross-hop topology, forms a directed set of hop points based on existing path relations, extracts the gate table identifiers of both upstream and downstream ends of each hop, the last time of gate edge change and the open-close sequence summary, and outputs the aligned view entries after conflict merging.

[0015] In a preferred embodiment, the aligned view entries contain the hop point identity, the version matching status of both ends, and the edge phase summary, and give three-level labels of ready, doubtful, and not ready, and the aligned view is landed on the data plane as the only external basis for judgment to avoid drift in the caliber.

[0016] In a preferred embodiment, an inactivated shadow gate table is generated in the data plane, and according to the consistent switching judgment result, the locked port and queue are selected and the switching token is generated, or the delayed review is performed, or the trigger is directly rejected, wherein the open-close events within the window are extracted from the dequeue log and the adjacent gate table, and the edge event sequence is reconstructed according to the window label.

[0017] In a preferred embodiment, the edge phase coverage is calculated from the edge event sequence, and the actual dequeue edge falls into the coverage level of the open segment of the opposite end to give three levels of high, medium, and low, and then the queuing and dequeue trajectories of the same service before and after the update window are tracked, and the truncated or split clues are marked, and the rule disambiguation is made for compliance shunting to form three levels of high, medium, and low of window through continuous degree.

[0018] In a preferred embodiment, the three levels of edge phase coverage and the three levels of window through continuous degree are input into a consistent domain mapper, which is mapped to a safe domain, a buffer domain, and a rejection domain in a two-dimensional parameter plane, and outputs a phase-through decision coefficient, and when the coefficient is high, the port and queue are switched to the locked state and the switching token is generated, when the coefficient is medium, the doubtful record is recorded and the delayed review is set, and when the coefficient is low, the trigger and locking are rejected.

[0019] In a preferred embodiment, the switching token is the only trigger, the shadow gate table is enabled simultaneously in the upstream and downstream, and the old gate table is kept to cover the update window to maintain uninterrupted dequeue and continuous circuit mapping, wherein after receiving the switching token, both ends enter the synchronous enable stage, and the dequeue scheduling adopts the open-close edges of the shadow gate table, and the more strict open-close combination is used in the window overlap area to avoid amplification and release.

[0020] In a preferred embodiment, the queue pointer only makes one-way advancement to eliminate the risk of backtracking, the port state keeps a traceable label during the overlap period, the conflict between the shadow gate table and the old gate table is decided by the shadow gate table first, and the old gate table automatically exits after the coverage expires.

[0021] A circuit resource dynamic matching system for different network exchanges, comprising:

[0022] Window marker distribution unit: distribute window markers in control plane, and aggregate gatetable version and adjacency digest of each node to generate alignment view as prerequisite of consistent switching:

[0023] Shadow table generation unit: generate inactive shadow gatetable in data plane, select lock port and queue according to consistent switching decision result, and generate switching token or perform delayed review or directly reject trigger:

[0024] Shadow table enabling unit: take switching token as the only trigger to enable shadow gatetable in upstream and downstream at the same time, and reserve old gatetable coverage update window to maintain continuous out-of-queue and circuit mapping:

[0025] Edge signal comparison unit: in the first valid window after enabling, return edge confirmation signal from downstream, compare actual out-of-queue and shadow gatetable in upstream, and if deviation is found, perform single backtracking and mark error jump point as subsequent correction basis:

[0026] Stability record generation unit: aggregate edge confirmation signal and alignment view to generate stability record, freeze window marker according to record solidification next batch update sequence and reservation time length rule, and complete this update.

[0027] The technical effects and advantages of the circuit resource dynamic matching method and system for different network switching of the present application are as follows:

[0028] The present application distributes window markers in control plane and generates alignment view, creates shadow gatetable in data plane, locks port and queue according to consistent switching decision, synchronously enables new table with switching token and reserves old table coverage, combines downstream edge confirmation and upstream deviation comparison to perform backtracking, aggregates signals to optimize update strategy, forms closed-loop circuit resource dynamic matching mechanism, effectively eliminates mapping instability in update window, ensures business out-of-queue continuity and network transmission reliability, and significantly improves transmission stability and configuration deployment efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 The flowchart of the circuit resource dynamic matching method for different network switching of the present application is shown.

[0030] Figure 2 The structural diagram of the circuit resource dynamic matching system for different network switching of the present application is shown. DETAILED DESCRIPTION

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Example 1: Figure 1 This invention provides a method for dynamic matching of circuit resources for switching between different networks, comprising:

[0033] S1: Mark and summarize the gating table version and adjacency summary of each node in the control plane distribution window, and generate an aligned view as a prerequisite for consistent switching;

[0034] S2 generates an inactive shadow gating table in the data plane, selects to lock the port and queue based on the consistency switching determination result and generates a switching token, or performs delayed review, or directly refuses to trigger;

[0035] S3 uses the switching token as the sole trigger, simultaneously enabling the shadow gating table in both upstream and downstream, while retaining the old gating table to overwrite the update window, thus maintaining uninterrupted dequeueing and continuous circuit mapping.

[0036] S4 returns an edge confirmation signal from downstream in the first valid window after activation. Upstream compares the actual outgoing queue with the shadow gating table. If a deviation is found, a single backtrack is executed and the error jump point is marked as a basis for correction.

[0037] S5 summarizes the edge confirmation signals and aligns the view to generate a stability record, and based on the record, it solidifies the update order and retention time rules for the next batch, freezes the window markers, and completes this update.

[0038] In different switching networks, critical services rely on gating time slots to maintain a stable correspondence between ports, queues, and time slots, ensuring consistent queuing and release order along the service path. However, during actual configuration updates, differences in the order of configuration propagation, brief delays in driver loading, and minor offsets in node local clocks can create a brief update window. This can lead to misalignment of gating table versions between adjacent nodes, disrupting the consistent mapping required for dynamic matching of circuit resources and causing short-term interruptions, minor out-of-order issues, or jitter accumulation. This problem directly impacts network reliability and service continuity. Therefore, it is necessary to eliminate version differences between old and new gating tables through window marking distribution and alignment view construction on the control plane, providing a prerequisite for subsequent consistent switching and ensuring a smooth recovery of the mapping from the transition period to normal.

[0039] Based on the mapping instability problem in the update window described in the background art, step S1 broadcasts the update intention and window mark to the participating nodes through the control plane, and aggregates the reported information to construct the alignment view, thereby laying the foundation for eliminating the open-closed edge misalignment across the hops. However, in actual execution, the generation process of the alignment view needs to be precisely defined to avoid conflicts and drifts in the version summary, and to ensure that the generated view serves as the only external basis for the data plane decision.

[0040] Specific processing technology logic of step S1:

[0041] The control plane broadcasts the update intention and window mark to the participating nodes, and the window mark is defined as a time interval for identifying the transition period of configuration update. After receiving the broadcast, the node reads the local gate table version, which is a unique identifier representing the current configuration state of the gate table; at the same time, it extracts the version summary of the adjacent port, which includes the gate table version information of the adjacent node. Subsequently, the node reports these information to the control plane.

[0042] The control plane forms a directed hop set based on the existing path relationship, which represents the directional links on the service path in the network, and each hop is composed of an upstream node and a downstream node. For each hop, the gate table identifier of the upstream and downstream ends is extracted, which is a string or a numerical value used to uniquely distinguish the gate table version; at the same time, the time of the last gate edge change is recorded, which is a timestamp representing the last modification time of the open-closed edge in the gate table; in addition, the open-closed sequence summary is extracted, which is a sequence describing the open and close sequence of the gate time slot.

[0043] In the conflict merging process, for each hop, if the gate table identifiers of the upstream and downstream ends are inconsistent, it is marked as a potential conflict; by comparing the time of the last gate edge change, the time difference value is calculated, which is defined as the difference between the change time of the downstream end and the change time of the upstream end, used to evaluate the sequence of version update. If the time difference value is greater than zero, it indicates that the downstream end changes later, which may cause the upstream dequeue to be ahead of time; if the time difference value is less than zero, it may cause lag. Combined with the open-closed sequence summary, the phase matching degree is calculated, which is defined as the intersection size of the open-closed sequence summary set of the upstream end and the open-closed sequence summary set of the downstream end divided by the union size of each other, representing the overlap ratio of the two sequences, dimensionless, with a value range of zero to one, used to quantify the similarity of the edge phase.

[0044] The output alignment view entry includes the identity of each hop, which is a unique identifier such as a link ID; the version match status of both ends, which is classified as match or mismatch based on whether the GATE table identifier is the same; and the edge phase digest, which is a combined on-off sequence summary. According to the time difference value and the phase matching degree, three levels of labels are given: ready, doubtful, and not ready. If the absolute value of the time difference is less than the time difference threshold and the phase matching degree is greater than the matching degree threshold, it is labeled as ready, where the time difference threshold is used to define the acceptable change time offset, and the matching degree threshold is used to define sufficient sequence overlap. If either condition is not met but the phase matching degree is greater than 0.5, it is labeled as doubtful. Otherwise, it is labeled as not ready. For example, the time difference threshold can be determined by simulating the average clock offset of the update window, such as measuring the clock difference of multiple nodes in a laboratory environment, and taking the upper bound of the 95% confidence interval as the threshold. The matching degree threshold can be based on the overlap statistics of the on-off sequence of historical traffic paths, and the minimum value that covers 90% of stable cases is selected.

[0045] Step S1 generates an alignment view containing three levels of labels by marking the broadcast window, summarizing the version and digest, forming a directed hop set and performing conflict merging, thereby providing accurate preconditions for version difference alignment within the update window, ensuring that circuit resources dynamically match to restore consistent mapping during the transition period, and avoiding short-term interruption and out-of-order problems.

[0046] Step S1 has generated an alignment view through the control plane and landed on the data plane, providing a unique external basis for version alignment and edge checking across hops, thereby supporting the foundation for eliminating version differences within the update window. However, before entering actual switching, further preparation of shadow GATE tables and consistency determination based on the alignment view and window markers are needed on the data plane to quantify misalignment risks and determine switching feasibility, avoiding mapping interruptions caused by blind triggering.

[0047] The specific processing technology logic of step S2 is as follows:

[0048] The data plane first generates an inactive shadow GATE table, which is a backup configuration created based on the new GATE table version and is in a non-execution state, used only for simulating switching effects. Subsequently, the consistent switching determination phase is entered, and the specified on-off events within the window markers are extracted from the out-of-log and adjacent GATE tables. These events include the time point records of port opening and closing. The edge event sequence is reconstructed according to the window markers, which is an ordered list arranging the occurrence time and type of all extracted events.

[0049] For the reconstructed edge event sequence, the edge phase coverage is calculated, which evaluates the coverage level of actual outgoing edges falling into the open segment at the opposite end. Specifically, for each hop, using the edge phase summary in the aligned view as reference, the coverage is defined as the total overlap of all upstream outgoing intervals and downstream open segments divided by the total length of upstream outgoing intervals, where the upstream outgoing interval refers to the difference between each outgoing open time and the corresponding close time at the upstream end, the downstream open segment refers to the difference between the corresponding open start time and end time at the downstream end, and the overlap represents the intersection length of these intervals in time, ensuring that the dimension is in time units, which quantifies the synchronization degree of upstream outgoing and downstream receiving. According to the value of the coverage, three levels of high, medium and low are given: if the coverage is greater than the upper threshold, it is high; if the lower threshold is less than the coverage and less than or equal to the upper threshold, it is medium; otherwise, it is low, where the upper threshold is used to define the high synchronization, and the lower threshold is used to define the minimum acceptable level. For example, the upper threshold can be selected by analyzing the outgoing log of the historical update window, selecting the percentile point when no interruption occurs when the coverage exceeds this value, such as the minimum value under 90% confidence level; the lower threshold is selected based on the simulation of the misalignment scenario, selecting the critical point when the coverage is lower than this value, i.e., the out-of-order occurs.

[0050] Next, the queuing and outgoing trajectories of the same service before and after the update window are tracked, which are the service packet sequence paths extracted from the outgoing log. The truncated or split clues are marked, which refer to instances of service packets being interrupted or diverted within the window due to edge misalignment. Subsequently, the compliance diversion is rule disambiguated, i.e., applying pre-defined rules to filter out normal diversion events and only keep abnormal clues, forming the window through continuity, which evaluates the integrity of the service trajectory. The specific calculation is 1 - (number of abnormal clues ÷ (total number of service packets within the window + number of compliance diversion)), which represents the complement of the abnormal proportion, dimensionless, with a value range of zero to one, ensuring the quantification of service continuity. According to the value of the continuity, three levels of high, medium and low are given: if the continuity is greater than the upper threshold, it is high; if the lower threshold is less than the continuity and less than or equal to the upper threshold, it is medium; otherwise, it is low, where the upper threshold is used to define the high continuity, and the lower threshold is used to define the minimum acceptable level. For example, the upper threshold can be obtained from the trajectory statistics of stable service paths, selecting the threshold when the jitter accumulation is zero when the continuity is higher than this value; the lower threshold is determined based on fault injection testing, determining the point when the continuity is lower than this value, i.e., causing interruption.

[0051] The three levels of edge in-phase coverage and the three levels of window through-continuity are input into a consistent domain mapper, which divides the two-dimensional parameter plane into regions: high-high combinations are mapped to a safe domain, high-medium or medium-high are mapped to a buffer domain, and the rest are mapped to a rejection domain. According to the mapping result, an in-phase through-decision coefficient is output: high for the safe domain, medium for the buffer domain, and low for the rejection domain. If the coefficient is high, the port and queue involved are switched to a locked state, which prohibits further configuration changes and generates a switching token, which is a unique authorization identifier used to trigger subsequent enablement. If the coefficient is medium, a doubtful entry is recorded, including specific misalignment clues, and a delayed review is set, which is a timed recalculation process with an interval based on the window marker length. If the coefficient is low, triggering and locking are rejected, and the current state is directly maintained to avoid risk amplification.

[0052] Step S2 realizes accurate quantification of misalignment risk in the update window and switching decision by generating a shadow gating table, calculating edge in-phase coverage and window through-continuity, and outputting a decision coefficient through a consistent domain mapper, thereby ensuring that circuit resources dynamically match to avoid mapping instability during the transition period, maintain business continuity, and provide a reliable token for subsequent enablement.

[0053] In the data plane, the shadow gating table is prepared and consistent switching is determined. First, edge in-phase coverage and window through-continuity are calculated, and then a consistent domain mapper outputs an in-phase through-decision coefficient. When the coefficient is high, the port and queue are locked and a switching token is generated. When the coefficient is medium, a delayed review is performed. When the coefficient is low, triggering is rejected.

[0054] Step S2 has generated a switching token in the data plane and locked the port and queue based on the in-phase through-decision coefficient, providing an authorization basis for safe enablement of the shadow gating table, thereby quantifying and controlling switching risk. However, when performing the switch, it is necessary to strictly trigger with the switching token to ensure that the upstream and downstream enable the shadow gating table at the same time, and bridge the transition by covering the update window with the old gating table to avoid out-of-order interruption and mapping disruption.

[0055] The specific processing technology logic of step S3 is as follows:

[0056] The switching token, which is generated only when the coefficient is high, is used as the only trigger and is derived from the decision output of step S2, and is used to verify switching authorization. The shadow gating table is enabled at the same time on the upstream and downstream, and the enablement process requires both end nodes to start the execution logic of the shadow gating table immediately after receiving the switching token.

[0057] The execution sequence is as follows: after receiving the switching token, the upstream end and the downstream end enter the synchronous activation phase, which ensures that both ends are activated at the same time through the exchange of confirmation messages to avoid edge misalignment caused by one-sided activation. The de-queue scheduling uses the opening and closing edges of the shadow gate table, which are the opening and closing time points defined in the shadow gate table and used to guide the release of service packets.

[0058] A stricter opening and closing combination is used in the window overlap area, which refers to the intersection of the opening and closing edges of the shadow gate table and the old gate table, and only allows release when both are open to avoid amplifying the release, i.e., to prevent accidental additional service packets from passing through and causing out-of-order. Specifically, for each time slot, the effective open duration is calculated, which is defined as the minimum value of the closing time of the shadow gate table minus the opening time of the shadow gate table and the closing time of the old gate table minus the opening time of the old gate table, and the overlapping part of the opening time to the closing time of the shadow gate table is taken. The calculation principle is to limit the release window to the most conservative range through minimum value operation and intersection processing to ensure that de-queue is only performed during the overlapping open period, thereby maintaining de-queue continuity and quantifying synchronization control.

[0059] The queue pointer only makes one-way advancement, which means that the processing index of service packets in the queue only moves forward and is not allowed to back up to eliminate the risk of looping, i.e., to prevent repeated or lost processing caused by circular processing. In specific implementation, the queue pointer starts from the current position and advances packet by packet to the end of the shadow gate table coverage, and if a conflict is encountered, the advancement is paused until the edges are aligned.

[0060] The port state keeps a traceable flag during the overlap period, which is a state bit that records the historical configuration of the port within the update window for subsequent backtracking. The conflict between the shadow gate table and the old gate table is resolved by the shadow gate table, i.e., when the opening and closing edges are inconsistent, the rules of the shadow gate table are used as the dominant, and the old gate table is only used as a supplement.

[0061] The old gate table automatically exits after the coverage expires, which is based on the end time of the window marker. When the time exceeds the window marker, the old gate table configuration is disabled, and only the shadow gate table is kept running to ensure continuous transition of circuit mapping to the new state.

[0062] Step S3 realizes uninterrupted de-queue and continuous circuit mapping in the update window by triggering the synchronous activation of the shadow gate table with the switching token, using the strict opening and closing combination and one-way queue advancement, and keeping the old gate table coverage during the overlap period, thereby bridging the transition period risk and ensuring the stability and continuity of critical services during configuration update.

[0063] Step S3 has been triggered by the switch token to synchronize the upstream and downstream shadow gate tables, and to maintain the dequeuing continuity through the old gate table coverage, thereby bridging the transition risk of the update window. However, after the switch is enabled, in order to verify the actual effect and capture the residual misalignment, the edge confirmation signal is returned by the downstream in the first valid window, and the upstream performs accurate comparison to determine whether to back off and mark the problem point.

[0064] The specific processing technique logic of step S4 is as follows:

[0065] In the first valid window after the enablement, the edge confirmation signal is sent by the downstream end, which contains the actual release of the first and last edges, i.e., the start and end time points of the service packet release observed by the downstream end, and the queue switching time summary, which is a timestamp set recording the moment when the queue switches from the old configuration to the new configuration.

[0066] After receiving the edge confirmation signal, the upstream end compares it with the window edges of the shadow gate table, which are the pre-defined open and close time boundaries in the shadow gate table. The specific comparison process involves calculating the time offset, which is defined as the absolute value of the actual release start edge minus the expected start edge of the shadow gate table, and then adding the absolute value of the actual release end edge minus the expected end edge of the shadow gate table. The deviation sum of the first and last edges is quantified by the sum of the absolute differences, and the overall misalignment degree is evaluated by accumulating the offset; at the same time, the queue switching time summary is compared with the corresponding time of the shadow gate table, and if there is a mismatch in the summary, it is marked as a potential truncation.

[0067] If continuous misalignment or truncation clues are observed, the clue refers to the offset of continuous multiple time slots exceeding a threshold value or the summary not matching, the threshold value such as the offset threshold value, for example, the 95% confidence upper limit of the out-of-order caused by the offset exceeding the value can be selected as the threshold value through historical switch log statistics to quantify the continuity interruption; in conjunction with the jump point entry of the alignment view, which provides the jump point identity and version matching state as a reference, a single backoff to the old gate table is immediately performed, which refers to the temporary recovery of the dequeuing scheduling of the old configuration.

[0068] At the same time, the error jump point is marked, which is a specific link point identified based on the jump point identity, recording the deviation details; the switch token and the lock state are revoked, which clears the authorization and unlocks the port and queue; the positioning record is output, which contains the error jump point, the offset, and the clue summary, facilitating subsequent review of the gate table or clock coordination.

[0069] If the comparison passes, i.e. the offset is less than the offset threshold and there is no truncation clue, the overlap flag on the port, which is derived from the overlap period status bit in step S3, is cleared and the shadow gate table is solidified to the running state, which means that the shadow gate table is permanently activated as the main configuration, and the transition is completed.

[0070] Step S4 realizes the timely capture and correction of the post-activation misalignment through the upstream comparison of the downstream edge confirmation signal, offset detection and single back-off mechanism, thereby providing a reliable basis for the stable ending of the update window, ensuring the dynamic matching of circuit resources to avoid persistent interruption, and supporting subsequent optimization in problem positioning.

[0071] Step S4 has completed offset detection and back-off processing through the upstream comparison of the downstream edge confirmation signal, and outputted the positioning record, which provides immediate feedback for the verification and correction of the update process, thereby ensuring the reliable solidification of the shadow gate table. However, to realize the closed-loop optimization of the entire update, it is necessary to further aggregate these signals and alignment views to generate a stability record, in order to derive the next batch of rules and freeze the window marker, completing this update.

[0072] The specific processing logic of step S5 is as follows:

[0073] The stability record is generated by aggregating the edge confirmation signal and the alignment view entry. First, the readiness ratio of the jump point level is calculated. The readiness ratio evaluates the proportion of jump points marked as ready in the alignment view. Specifically, the readiness ratio is defined as the number of jump points marked as ready in the alignment view divided by the total number of jump points, which quantifies the overall readiness through simple proportion, is dimensionless, and has a value range of zero to one, reflecting the global success rate of version alignment. At the same time, the level distribution of the number of back-offs and the number of passes is counted. The distribution divides the number of back-offs into low (zero to one), medium (two to three), and high (more than three) grades. By counting the back-off instances in the edge confirmation signal and corresponding to the pass instances (success of no offset comparison), the distribution count is formed.

[0074] According to the stability record, the update order of the next batch is derived. The jump points with high stability are prioritized, i.e. the jump points with a readiness ratio greater than the order threshold are placed first. The order threshold may be based on the records of historical update batches, selecting the critical value of zero back-off rate when the readiness ratio is higher than the value as the starting point of optimization. The jump points prone to misalignment are extended to overlap coverage. The prone-to-misalignment refers to the jump points with a high proportion of high-grade back-off in the distribution. The extended coverage duration is 1.5 times the original window marker length. The coverage threshold can be obtained from simulation tests, selecting the statistical upper limit of amplified jitter when the proportion of high-grade exceeds the value.

[0075] The stable passing jump point is shortened and the order is pre-arranged, the stable passing refers to the jump point with high passing frequency distribution (passing rate greater than 0.8) and low backtracking, the coverage is shortened to 0.7 times of the original length, and the order is arranged at the front end of the update sequence to speed up the processing.

[0076] Finally, the freeze window mark is marked, the freeze refers to setting the window mark to an unmodifiable state, and ending the current update to form a traceable update closed loop, the closed loop includes a complete record chain from S1 to S5, and a reusable deployment recipe, the recipe is based on the stability record rule template for subsequent batches to be directly applied.

[0077] The step S5 realizes the closed loop optimization of the update process by summarizing the edge confirmation signal and the alignment view to generate the stability record containing the proportion and the distribution, deriving the optimization rule and freezing the window mark, thereby providing traceable feedback and reuse strategy for continuous deployment of the switching network, and ensuring that the circuit resource dynamic matching maintains high efficiency and stability in multiple batches of updates.

[0078] Embodiment 2: Figure 2 A circuit resource dynamic matching system for different network switching is provided, comprising:

[0079] The window mark distribution unit distributes the window mark in the control plane, and summarizes the gate table version and the adjacency digest of each node to generate the alignment view as the precondition of consistent switching.

[0080] The shadow table generation unit generates the inactivated shadow gate table in the data plane, selects the lock port and the queue according to the consistent switching judgment result, and generates the switching token, or performs the delayed review, or directly rejects the trigger.

[0081] The shadow table enabling unit enables the shadow gate table in the upstream and the downstream at the same time with the switching token as the only trigger, and retains the old gate table coverage update window to maintain the uninterrupted dequeuing and continuous circuit mapping.

[0082] The edge signal comparison unit returns the edge confirmation signal from the downstream in the first effective window after enabling, and compares the actual dequeuing and the shadow gate table in the upstream, and if a deviation is found, a single backtracking is performed and an error jump point is marked as a basis for subsequent correction.

[0083] The stability record generation unit generates the stability record by summarizing the edge confirmation signal and the alignment view, solidifies the update sequence and the retention time rule of the next batch according to the record, freezes the window mark, and completes the current update.

[0084] The above formulas are dimensionless and the numerical values are calculated, the formula is obtained by collecting a large amount of data to simulate a formula of the latest real situation, and the preset parameters in the formula are set by the person skilled in the art according to the actual situation.

[0085] It should be noted that the system of the present application can be deployed in the device itself to realize embedded application, or can be run on PC or other terminal with user interface, so as to meet various hardware environment and use requirements.

[0086] The above only describes certain exemplary embodiments of the present application by way of illustration, and it is needless to say that the described embodiments can be modified in various ways without departing from the spirit and scope of the present application by those skilled in the art. Therefore, the above figures and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present application.

[0087] It should be noted that in this document, relational terms such as first and second and the like can merely be used to distinguish one entity or action from another, without necessarily requiring or implying that there is any such actual relationship or order between or among the entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0088] The above is merely a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for dynamic matching of circuit resources exchanged by different networks, characterized in that, The method comprises the steps of: In the control plane, distribute window markers and aggregate the gatetable versions and adjacency summaries of each node to generate an alignment view as a prerequisite for consistent switching; In the data plane, generate an inactive shadow gatetable, select locked ports and queues and generate switching tokens according to the consistent switching decision, or perform a delayed review, or directly reject triggering; Extract open and close events within the window from the dequeuing log and adjacency gatetable, and reconstruct the edge event sequence according to the window marker; Calculate the edge in-phase coverage from the edge event sequence, and give three levels of high, medium and low according to the coverage level of the actual dequeuing edge falling into the open segment of the opposite end, then track the queuing and dequeuing trajectory of the same service before and after the update window, mark the truncated or split clues, and rule disambiguate the compliance shunting to form three levels of high, medium and low of window through-continuity; Input the three levels of edge in-phase coverage and window through-continuity into a consistent domain mapper, map to a safe domain, a buffer domain and a rejection domain in a two-dimensional parameter plane, and output an in-phase through-cutoff coefficient, if the coefficient is high, switch the involved ports and queues to the locked state and generate a switching token, if the coefficient is medium, record the suspect and set a delayed review, and if the coefficient is low, reject triggering and locking; Use the switching token as the only trigger to enable the shadow gatetable simultaneously in the upstream and downstream, and keep the old gatetable covered in the update window to maintain uninterrupted dequeuing and circuit mapping continuity; In the first effective window after enabling, return the edge confirmation signal from the downstream, compare the actual dequeuing with the shadow gatetable in the upstream, and if a deviation is found, perform a single backtracking and mark the error jump point as a correction basis; Generate a stability record by aggregating the edge confirmation signal and the alignment view, and according to the record, solidify the update order and retention time rule of the next batch, freeze the window marker, and complete this update.

2. The circuit resource dynamic matching method for different network exchanges according to claim 1, characterized in that: In the control plane, distribute window markers and aggregate the gatetable versions and adjacency summaries of each node to generate an alignment view as a prerequisite for consistent switching, wherein the control plane broadcasts the update intention and window marker to the participating nodes, and the nodes read the local gatetable version and version summary of the adjacent port and report them.

3. The circuit resource dynamic matching method for different network exchanges according to claim 2, characterized in that: The control plane generates an alignment view of the cross-hop topology according to the report, forms a directed hop point set based on the existing path relationship, extracts the gatetable identifier, the last gatetable edge change time and the open and close sequence summary of the upstream and downstream of each hop, and outputs the alignment view entry after conflict merging.

4. The circuit resource dynamic matching method for different network exchanges according to claim 3, characterized in that: The alignment view entry contains the hop point identity, the two-end version matching state and the edge phase summary, and gives three levels of ready, suspect and not ready, and the alignment view is landed to the data plane as the only external basis for judgment to avoid drift.

5. The circuit resource dynamic matching method for different network exchanges according to claim 1, characterized in that: The shadow gating table is enabled simultaneously in upstream and downstream with the switch token as the only trigger, and the old gating table coverage update window is retained to maintain continuous circuit mapping and uninterrupted dequeuing. After receiving the switch token, both ends enter the synchronous enabling stage. The dequeuing schedule uses the opening and closing edges of the shadow gating table. In the window overlap area, a stricter opening and closing combination is used to avoid amplification and release.

6. The circuit resource dynamic matching method for different network exchanges according to claim 5, characterized in that: The queue pointer only advances in one direction to eliminate the risk of backtracking. The port state is marked as retroactive during the overlap period. The conflict between the shadow gating table and the old gating table is resolved by the shadow gating table. After the coverage expires, the old gating table automatically exits.

7. A system for dynamic matching of circuit resources for different network exchanges for implementing the method for dynamic matching of circuit resources for different network exchanges according to any one of claims 1 to 6, characterized in that It includes: The window marker distribution unit distributes window markers in the control plane, aggregates the gating table versions and adjacency summaries of each node, generates an alignment view, and serves as a prerequisite for consistent switching. The shadow table generation unit generates an inactive shadow gating table in the data plane, selects the locked ports and queues based on the consistent switching decision result, generates a switch token, or performs a delayed review, or directly rejects the trigger. The shadow table enabling unit enables the shadow gating table simultaneously in upstream and downstream with the switch token as the only trigger, and retains the old gating table coverage update window to maintain continuous circuit mapping and uninterrupted dequeuing. The edge signal comparison unit returns the edge confirmation signal in the first valid window after enabling from the downstream. The upstream compares the actual dequeuing with the shadow gating table. If a deviation is found, a single backtracking is performed and an error jump point is marked as a basis for subsequent correction. The stability record generation unit generates a stability record based on the edge confirmation signal and the alignment view, solidifies the update sequence and retention time rules for the next batch, freezes the window marker, and completes the update.

Images