An enterprise data collaborative scheduling method and system for cross-platform integration

Abstract

The application discloses a kind of enterprise data collaborative scheduling methods and systems for cross-platform integration, specifically relates to enterprise data scheduling and data disaster control field, including obtaining the task execution record of each business platform, analyzing cross-platform data calling sequence, constructing the scheduling path set consisting of data processing node, and calculating the dependence intensity of each path as evaluation value;The application constructs the data processing path structure with task process chain as the core, integrates target state configuration, state snapshot generation and state verification record construction mechanism, and introduces dynamic evaluation and disaster recovery strategy based on path stability index, to solve the problem of scheduling interruption and disaster recovery failure caused by inconsistent data state and missing state recovery process in the existing technology in multi-platform heterogeneous system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of enterprise data scheduling and data disaster recovery control, and more particularly to an enterprise data collaborative scheduling method and system for cross-platform integration. BACKGROUND

[0002] In the cross-platform data collaborative scheduling of modern enterprises, how to ensure the flow and scheduling of data between different platforms becomes particularly difficult. The existing technologies mostly rely on traditional backup and recovery schemes, such as local backup, cloud storage redundancy, etc., but these schemes have obvious limitations in cross-platform environments. First, data synchronization problems occur frequently. Network delays, protocol incompatibilities, and system architecture differences between platforms result in data that cannot be synchronized in a timely or complete manner, increasing the risk of inconsistency during data recovery. Second, traditional disaster recovery techniques focus on the recovery of a single platform or a single system, making it difficult to handle complex scenarios that may arise during multi-platform collaborative operations, and unable to ensure that data can be consistently recovered in a global range in the event of system failure or emergencies.

[0003] Therefore, how to build a reliable data disaster recovery mechanism in the complex environment of cross-platform data collaborative scheduling to ensure that data between multiple platforms can be quickly and accurately recovered in the event of failure or emergencies, and to maintain data consistency during the recovery process has become the core problem. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide an enterprise data collaborative scheduling method and system for cross-platform integration, which builds a data processing path structure with a task process chain as the core, integrates target state configuration, state snapshot generation, and state verification record construction mechanisms, and introduces a dynamic evaluation and disaster recovery reconstruction strategy based on path stability index, to solve the problem of scheduling interruption and disaster recovery failure caused by inconsistent data states and missing state recovery processes in multi-platform heterogeneous systems in the prior art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical scheme: an enterprise data collaborative scheduling method for cross-platform integration, comprising:

[0006] S1, obtaining task execution records of each business platform, analyzing cross-platform data call order, building a scheduling path set composed of data processing nodes, and calculating the dependency strength of each path as an evaluation value;

[0007] S2, selecting paths with evaluation values higher than a threshold from the scheduling path set, setting the target state, error range, and recovery priority of each data processing node, and generating a task process chain according to the data processing node call order and platform information;

[0008] S3, before the task starts, call the state monitoring interface of each business platform, obtain the structured state data of each data processing node in the task process chain in the current period as the current state snapshot, and compare it with the target state of the corresponding node to generate a state check record;

[0009] S4, arrange the current state snapshot and the state check record according to the calling sequence of the data processing node into a task snapshot set, which is used for data disaster recovery judgment and state recovery;

[0010] S5, during task execution, compare the current state snapshot of each data processing node with the state check record in the task snapshot set according to the task process chain in real time, if the states are consistent, the data processing node continues to execute; if the state is abnormal, the current path is stopped, and each node state is recovered node by node according to the snapshot record;

[0011] S6, after the task is completed, update the evaluation value of each path in the scheduling path set according to whether the path has experienced a recovery operation; mark the path that triggers recovery twice in a row as an adjustment object, and reset the task process chain parameters of the path; keep the current setting for the path that does not trigger recovery.

[0012] In a preferred embodiment, in S1, the data processing nodes involved in each task are obtained by reading the task execution records of each business platform, and the data processing nodes are arranged into a task path sequence according to the execution order of the tasks;

[0013] The data processing nodes contained in each task in the task path sequence are grouped into a node sequence according to the calling sequence, and adjacent data processing nodes are connected to form a node connection relationship, all node connection pairs are extracted, and output as a directed edge set;

[0014] For each node connection pair in the directed edge set, the frequency of its occurrence in the task path sequence is counted, weight calculation is performed based on the frequency, and the calculation result is attached to the corresponding node connection pair to form a weighted directed edge set;

[0015] The data processing node connection pairs that can be traversed in the weighted directed edge set are combined into node paths, the weight values of each edge are accumulated for each path to obtain the dependence strength value of the path; all paths and their dependence strength form a scheduling path set, and the dependence strength value is used as the evaluation value of each path.

[0016] In a preferred embodiment, in S2, the evaluation value distribution of each path is obtained by reading the dependence strength value corresponding to each path in the scheduling path set, then the evaluation value interval division operation is performed, the high-risk path segment is identified according to the set dynamic threshold, the path with an evaluation value higher than the threshold is screened out, and the candidate path set is output;

[0017] By analyzing the data processing node call sequence of each path in the candidate path set, the data processing nodes appearing in the path in turn are identified, and the business platform identifiers corresponding to each node are extracted, and the data processing node information with platform identifier in sequence is outputted;

[0018] The data processing node information is inputted into the target state setting process, and a three-stage configuration operation is performed, including setting the data processing node target state based on platform historical task setting data, setting the error tolerance range according to the data processing node sensitivity, and allocating the recovery priority according to the path interruption history, and outputting the node information sequence containing complete configuration parameters;

[0019] The node information sequence containing complete configuration parameters is spliced into a task process chain according to the path order, wherein each data processing node contains node identifier, platform identifier, target state, error range and recovery priority.

[0020] In a preferred embodiment, in S3, by traversing each data processing node in the task process chain, the state monitoring interface of the corresponding business platform is called to collect the structured state data of each data processing node in the current running period, and the structured state data is recorded in sequence according to the order of the data processing node in the task process chain, and a current state snapshot is generated;

[0021] The current state snapshot is compared with the target state sequence set in the task process chain at the field level, and the state deviation value and deviation direction of each data processing node are calculated respectively. Before field-level comparison, a field normalization function is called in advance to uniformly map the structured state fields of cross-platform data processing nodes to solve the difference in data expression between platforms. The deviation value is calculated by the numerical difference between the normalized field current value and the target value, and the deviation direction is determined according to the difference value sign;

[0022] The state deviation value of each data processing node is compared with the corresponding set error range in absolute value, if the absolute value of the state deviation value is greater than the error range, the data processing node is marked as an out-of-limit state, and the abnormal type is marked according to the deviation direction; if the state deviation value is within the error range, it is marked as normal state; if the state data collection fails, it is marked as undetermined state. For the nodes marked as undetermined state, the delay judgment mechanism will be triggered in the subsequent execution process of the scheduling path, and will not be included in the state consistency comparison range, to ensure the continuity and stability of the whole scheduling chain;

[0023] According to the data processing node call sequence in the task process chain, the state mark results of each node are combined in turn to form a state comparison label sequence, and the label sequence is recorded together with the current state snapshot and the target state sequence as a state verification record.

[0024] In a preferred embodiment, in S5, by sequentially traversing each data processing node in the task process chain, the current state snapshot of each data processing node in the current task period is compared with the corresponding state check record in the task snapshot set at the field level, and the judgment flag of each comparison result is recorded in real time during the comparison process;

[0025] The data processing nodes with consistent field-level comparison results are added to the continue execution queue of the current scheduling path in the order of the task process chain, and the comparison operation of the subsequent nodes is continuously performed. If all node states are consistent, the task process chain completes data scheduling according to the original path.

[0026] If the state check record of any data processing node is marked as inconsistent during the comparison process, the execution process of the current scheduling path is immediately stopped, the current state snapshot content of each inconsistent node is extracted from the task snapshot set in sequence, and the structured state data in the snapshot is loaded to the corresponding data processing node.

[0027] After the state of the data processing node is restored, the state consistency comparison result record of the scheduling path during task execution is updated in real time, and the path with two or more consecutive data processing node state restorations is marked as a high-risk path.

[0028] In a preferred embodiment, in S5, by sequentially traversing each data processing node in the task process chain, the current state snapshot of each data processing node in the current task period is compared with the corresponding state check record in the task snapshot set at the field level, and the judgment flag of each comparison result is recorded in real time during the comparison process;

[0029] The data processing nodes with consistent field-level comparison results are added to the continue execution queue of the current scheduling path in the order of the task process chain, and the comparison operation of the subsequent nodes is continuously performed. If all node states are consistent, the task process chain completes data scheduling according to the original path.

[0030] If the state check record of any data processing node is marked as inconsistent during the comparison process, the execution process of the current scheduling path is immediately stopped, the current state snapshot content of each inconsistent node is extracted from the task snapshot set in sequence, and the structured state data in the snapshot is loaded to the corresponding data processing node.

[0031] After the state of the data processing node is restored, the state consistency comparison result record of the scheduling path during task execution is updated in real time, and the path with two or more consecutive data processing node state restorations is marked as a high-risk path.

[0032] In a preferred embodiment, an enterprise data collaborative scheduling system for cross-platform integration includes a path extraction module, a path screening module, a state acquisition module, an information integration module, an exception handling module, and a path updating module.

[0033] The path extraction module obtains task execution records of each business platform, parses cross-platform data calling sequences, constructs a scheduling path set composed of data processing nodes, and calculates the dependency strength of each path as an evaluation value.

[0034] The path screening module screens paths with evaluation values higher than a threshold from the scheduling path set, sets target states, error ranges, and recovery priorities of each data processing node, and generates a task process chain according to data processing node calling sequences and platform information.

[0035] The state acquisition module obtains structured state data of each data processing node in the current period in the task process chain before task startup by calling state monitoring interfaces of each business platform, as a current state snapshot, and compares it with the target state of the corresponding node to generate a state verification record.

[0036] The information integration module arranges the current state snapshot and the state verification record into a task snapshot set according to the data processing node calling sequence, which is used for data disaster judgment and state recovery.

[0037] The exception handling module compares the current state snapshot of each data processing node with the state verification record in the task snapshot set in real time during task execution, and if the states are consistent, the node continues to execute; if the state is abnormal, the current path is stopped, and each node state is recovered node by node according to the snapshot record.

[0038] The path updating module updates the evaluation value of each path in the scheduling path set according to whether the path has undergone a recovery operation after task completion; marks paths that have triggered recovery twice as adjustment objects and re-sets their task process chain parameters; and keeps the current settings for paths that have not triggered recovery for continuous scheduling.

[0039] Technical effects and advantages of the present application:

[0040] 1、The present application constructs a current state snapshot and a state verification record before task startup, and compares the state differences of data processing nodes in real time during task execution, and makes fine-grained judgments based on error ranges and deviation directions, and when an abnormal state is found, the path is stopped and the node state is recovered according to the snapshot, thereby ensuring that cross-platform data can be quickly, accurately and consistently recovered when a fault occurs, solving the problem of inconsistent data when recovering in the prior art under multiple platforms.

[0041] 2、The application forms a weighted directed edge set by analyzing the connection relationship and appearance frequency of the data processing node in each task execution record, and calculates the path dependence strength value, constructs the scheduling path set and path evaluation value, realizes the accurate measurement of the cross-platform data scheduling dependence degree, and provides an evaluation basis for subsequent disaster recovery strategy;

[0042] 3、The application refines to the deviation value and deviation direction by comparing the current state and target state at the field level, and judges in combination with the set error range, identifies the slight state anomaly and locates the abnormal node, realizes the sensitive response and accurate identification of the state change of the multi-platform node, and improves the real-time performance of the disaster recovery processing;

[0043] 4、The scheme further judges whether the deviation directions of the nodes with label conflicts are consistent, and marks as a state to be reviewed if the deviation directions are inconsistent, avoids the misjudgment recovery operation caused by transient fluctuation, and improves the stability of the recovery logic;

[0044] 5、The scheme records whether the state recovery is triggered in the task execution, constructs a stability index based on the number of continuous triggering times, periodically evaluates the path, and then dynamically adjusts the task process chain parameters and path scheduling priority, realizes the cross-period risk monitoring. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 The method step flowchart of the application.

[0046] Figure 2 The system module diagram of the application.

[0047] Figure 3 The scheduling path construction flowchart of the application.

[0048] Figure 4 The task snapshot and state verification flowchart of the application.

[0049] Figure 5 The task execution and disaster recovery processing flowchart of the application.

[0050] Figure 6 The path evaluation value and stability updating flowchart of the application. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application.

[0052] The accompanying drawings are referred to in Figures 1-6 The embodiment of the present application is a cross-platform integrated enterprise data collaborative scheduling method, comprising:

[0053] S1, obtain the task execution record of each business platform, parse the cross-platform data calling sequence, construct a scheduling path set composed of data processing nodes, and calculate the dependence strength of each path as an evaluation value;

[0054] S2, select the path with an evaluation value higher than the threshold from the scheduling path set, set the target state, error range and recovery priority of each data processing node, and generate a task process chain according to the data processing node calling sequence and platform information;

[0055] S3, before the task is started, call the state monitoring interface of each business platform to obtain the structured state data of each data processing node in the current period in the task process chain as the current state snapshot, and compare it with the target state of the corresponding node to generate a state verification record;

[0056] S4, arrange the current state snapshot and the state verification record according to the data processing node calling sequence into a task snapshot set, which is used for data disaster recovery judgment and state recovery;

[0057] S5, during task execution, real-time compare the current state snapshot of each data processing node with the state verification record in the task snapshot set according to the task process chain, if the states are consistent, continue to execute; if the state is abnormal, stop the current path, and recover the state of each node according to the snapshot record;

[0058] S6, after the task is completed, update the evaluation value of each path in the scheduling path set according to whether the path has experienced a recovery operation; mark the path that triggers recovery twice in a row as an adjustment object, and reset its task process chain parameters; keep the current settings for the path that does not trigger recovery, which is used for continuous scheduling.

[0059] In S1, by reading the task execution record of each business platform, the data processing nodes involved in each task are obtained, and the task path sequence is arranged according to the execution order of the task;

[0060] The data processing nodes contained in each task in the task path sequence are grouped into node sequences according to the calling sequence, and the adjacent data processing nodes are connected to form a node connection relationship, all node connection pairs are extracted, and output as a directed edge set;

[0061] For each node connection pair in the directed edge set, count its frequency in the task path sequence, perform weight calculation based on the frequency, and append the calculation result to the corresponding node connection pair to form a weighted directed edge set;

[0062] The data processing nodes that can be traversed in the set of weighted directed edges are connected to form node paths. The weight values ​​of each edge in each path are accumulated to obtain the dependency strength value of the path. All paths and their dependency strengths are combined to form a set of scheduling paths, and the dependency strength value is used as the evaluation value of each path.

[0063] It should be noted that in the formula structure involved in this scheme, dimensionless terms can be used as proportional or structural adjustment factors. When combined with quantities with units, they only play a role in numerical scaling and do not introduce new physical dimensions. Therefore, they will not change or confuse the overall unit system. This combination of "dimensionless terms and terms with units" can be understood as a composite structural expression commonly used in mathematical physics modeling. It conforms to the principle of dimensional consistency and has a clear physical interpretation basis.

[0064] Secondly, in the formula structure of this scheme, if multiple variables with different physical units are involved, including but not limited to time, mass or energy variables, their joint appearance is to express the collaborative modeling relationship of multiple physical mechanisms. Each variable can form a unified structure through function mapping, ratio combination or normalization adjustment, with clear units and clear meaning. The overall expression conforms to the principle of dimensional consistency and the conventional formula of engineering modeling.

[0065] In this solution, constants, weights, adjustment factors, threshold parameters, proportional coefficients, etc., are all adjustable control parameters for different application environments. Their values ​​depend on the target equipment configuration, data input characteristics, and performance optimization goals. During the implementation phase, they are set to converge within a reasonable range through model verification, performance constraints, or engineering calibration. Although these parameters do not have a unique preset value, they have clear adjustment logic and calculation paths. They belong to the deterministic setting process in engineering implementation. The purpose of this setting is to ensure that the solution is both universally adaptable and reproducible and operable, without affecting its technical clarity and feasibility.

[0066]

[0067] C u→v =1+log2(1+R) u,v )

[0068]

[0069] Where P is the scheduling path to be evaluated; ε is the set of directed edges appearing in all historical task paths; (u→v) is the directed connection from data processing node u to data processing node v; F u→v W represents the total frequency of the edge (u→v) across all task paths. u→v R represents the frequency-normalized weight of the edge (u→v); u,vR u,v represents the complexity of platform heterogeneous interaction; C u→v is the platform coupling factor of edge (u→v), the platform coupling factor is obtained by taking the logarithm of R u,v , reflecting the relative complexity of the edge in cross-platform connection; L P is the number of node connection edges contained in the path P, L P is used to measure the structural length of the path; Φ(P) is the path connectivity reduction function, which performs length adjustment on the path score value; G v is the role level weight of node v, G v is used to reflect the core degree of the node in the task; S P is the final dependence strength score of the path P, which is the total value of the weighted score of each edge in the path structure multiplied by the connectivity adjustment function, which is used as the evaluation basis for subsequent scheduling screening and sorting.

[0070] Further, first, the frequency normalization formula W u→v measures the relative importance of node connection in the global path structure, then the platform coupling factor C u→v strengthens the distinction of the interaction complexity between heterogeneous platforms, then the node role level weight G v injects the functional difference of the node in the task flow, finally all the weighted connection scores are aggregated, and the path length L P adjusts the overall score density through the path reduction function Φ(P), forming the path dependence strength value S P , which is used to sort and schedule the stability and priority of the path, so as to screen out the optimal scheduling path in the complex cross-platform scenario.

[0071] In S2, by reading the dependence strength value corresponding to each path in the scheduling path set, the evaluation value distribution of each path is obtained, then the evaluation value interval division operation is performed, the high-risk path segment is identified according to the set dynamic threshold, the path whose evaluation value is higher than the threshold is screened out, and the candidate path set is output;

[0072] By analyzing the data processing node call sequence of each path in the candidate path set, the data processing nodes appearing in the path in turn are identified, and the business platform identifiers corresponding to each node are extracted, and the data processing node information with platform identifier arranged in order is output;

[0073] The data processing node information is input into the target state setting process, and a three-stage configuration operation is performed, including setting a target state of the data processing node based on platform historical task setting data, setting an error tolerance range according to data processing node sensitivity, and allocating a recovery priority according to path interruption history, and outputting a node information sequence containing complete configuration parameters;

[0074] The node information sequence containing complete configuration parameters is spliced into a task process chain in path order, wherein each data processing node contains a node identifier, a platform identifier, a target state, an error range, and a recovery priority, for subsequent state verification and disaster recovery processing in the scheduling task.

[0075] In S3, by traversing each data processing node in the task process chain, the state monitoring interface of the corresponding business platform is called to collect the structured state data of each data processing node in the current running period, and the structured state data is recorded in sequence according to the order of the data processing node in the task process chain to generate a current state snapshot;

[0076] The current state snapshot and the target state sequence set in the task process chain are compared at the field level, and the state deviation value and deviation direction of each data processing node are calculated respectively. Before the field-level comparison, a field normalization function is called in advance to uniformly map the structured state fields of cross-platform data processing nodes to solve the difference in data expression between platforms. The deviation value is calculated from the numerical difference between the current value and the target value of the normalized field, and the deviation direction is determined according to the difference value;

[0077] The absolute value of the state deviation value of each data processing node is compared with the corresponding set error range. If the absolute value of the state deviation value is greater than the error range, the data processing node is marked as an out-of-limit state, and the abnormal type is marked according to the deviation direction. If the state deviation value is within the error range, it is marked as a normal state. If the state data collection fails, it is marked as an undeterminable state. For the nodes marked as undeterminable state, a delayed judgment mechanism will be triggered in the execution process of the subsequent scheduling path, and will not be included in the state consistency comparison range to ensure the continuity and stability of the overall scheduling chain;

[0078] According to the calling order of the data processing nodes in the task process chain, the state marking results of each node are combined in sequence to form a state comparison label sequence, and the label sequence is recorded together with the current state snapshot and the target state sequence as a state verification record for state consistency comparison and abnormality judgment in the scheduling task execution process.

[0079] In S4, by reading the order of each data processing node in the task process chain, the corresponding current state snapshot and state verification record are extracted in sequence, and the two types of state information are paired and combined according to the node calling order to form a state comparison sequence;

[0080] The state comparison sequence is divided into several sliding comparison intervals according to the arrangement order of the data processing nodes, each interval contains a fixed number of continuous state pairs, and then the matching consistency of the current state and the state check mark is compared in each comparison interval, and all mark conflict positions are recorded;

[0081] For the data processing node corresponding to the identified conflict position, the target state and error range are continuously read, and the deviation direction of the current state relative to the target state is calculated; if the deviation direction is inconsistent with the deviation direction recorded in the state check record, the state mark of the node in the state comparison sequence is replaced with the state to be reviewed;

[0082] All state comparison items not marked as to be reviewed are recombined according to the node call order in the task process chain to construct a task snapshot set, and the task snapshot set and the conflict node index information are output together to support path stability judgment and disaster recovery strategy preprocessing in the scheduling task execution process.

[0083] In S5, by sequentially traversing each data processing node in the task process chain, the current state snapshot of each data processing node in the current task period is compared with the corresponding state check record in the task snapshot set at the field level, and the judgment flag of each comparison result is recorded in real time during the comparison process;

[0084] The data processing nodes with consistent field-level comparison results are added to the continue execution queue of the current scheduling path according to the task process chain order, and the comparison operation of the subsequent nodes is continuously performed, and if all node states are consistent, the task process chain completes data scheduling according to the original path;

[0085] If the state check record of any data processing node is marked as inconsistent during the comparison process, the execution process of the current scheduling path is immediately stopped, the current state snapshot content of each state inconsistent node is extracted from the task snapshot set in sequence, and the structured state data in the snapshot is loaded to the corresponding data processing node;

[0086] After the state of the data processing node is restored, the state consistency comparison result record of the scheduling path in the task execution process is updated in real time, and the path with two or more consecutive data processing node state restorations is marked as a high-risk path, which provides a judgment basis for subsequent task process chain parameter resetting and scheduling path evaluation updating.

[0087] In S6, after each scheduling path completes the scheduling task, the data processing nodes in the task process chain are traversed, the corresponding state check record is extracted, it is judged whether the node state recovery operation triggered based on the state deviation value exists in the scheduling path, and the recovery trigger result is taken as the risk label of the path scheduling history;

[0088] Based on the risk label recorded in the scheduling history, the recovery trigger frequency of each scheduling path in the current and previous scheduling period is calculated, and a path scheduling stability index is constructed to reflect the consistency deviation trend of the path in the last two execution processes;

[0089] If the stability index of the scheduling path meets the judgment condition of continuous trigger recovery, the path is marked as a risk rising path, and based on the current task process chain parameters, the target state, error range and recovery priority of the data processing node are reconfigured to form an updated task process chain;

[0090] If the stability index of the scheduling path is lower than the abnormal intervention threshold, and it does not trigger any recovery operation in the current period, the keep evaluation value is kept unchanged and used for subsequent scheduling priority sorting, and the priority is reordered according to the stability index of each path in the current period for subsequent continuous scheduling execution.

[0091] An enterprise data collaborative scheduling system for cross-platform integration includes a path extraction module, a path screening module, a state acquisition module, an information integration module, an abnormality processing module and a path updating module;

[0092] The path extraction module obtains the task execution records of each business platform, analyzes the cross-platform data calling sequence, constructs a scheduling path set composed of data processing nodes, and calculates the dependence strength of each path as an evaluation value;

[0093] The path screening module screens the paths with evaluation values higher than the threshold from the scheduling path set, sets the target state, error range and recovery priority of each data processing node, and generates a task process chain according to the data processing node calling sequence and platform information;

[0094] The state acquisition module acquires the structured state data of each data processing node in the current period in the task process chain as a current state snapshot by calling the state monitoring interface of each business platform before task starting, and compares the fields of the target state of the corresponding node to generate a state check record;

[0095] The information integration module arranges the current state snapshot and the state check record into a task snapshot set according to the data processing node calling sequence, which is used for data disaster recovery judgment and state recovery;

[0096] The abnormality processing module compares the current state snapshot of each data processing node with the state check record in the task snapshot set according to the task process chain in task execution, and if the states are consistent, the node continues to execute; if the state is abnormal, the current path is stopped, and the state of each node is recovered node by node according to the snapshot record;

[0097] The path updating module updates the evaluation value of the path in the scheduling path set according to whether the path has experienced a recovery operation after the task is completed; the path that triggers recovery for two consecutive times is marked as an adjustment object, and the task process chain parameters thereof are reset; the path that does not trigger recovery keeps the current setting and is used for continuous scheduling.

[0098] Further, the whole process formed by the scheme covers the whole process from task record analysis, path dependence calculation, state comparison and recovery to final path stability feedback. First, in the early stage of task execution, the scheme extracts task execution records from each business platform, and models the data processing nodes involved in the task as paths. The directed edge set is formed by building the connection relationship of the nodes, and the dependence strength between cross-platform tasks is quantified by accumulating the frequency weight of these connection pairs. The number of edges in the path is defined as the path length, and the accumulated total weight value is used as the evaluation value of the path, thereby forming the scheduling path set, ensuring that the subsequent process has a clear scheduling target basis;

[0099] In the middle stage after path identification, the scheme performs a three-stage state parameter configuration process based on the sequence of data processing nodes in the candidate path and the platform identifier, and determines the target state, error tolerance range and recovery priority in turn. These configurations are organized in a structured form as a task process chain. Then, state snapshot collection and real-time check operations are performed before and during task startup, respectively. The actual state is compared with the set target at the field level, and the deviation degree and direction of each node are marked. Local conflict detection is performed by sliding interval, and the inconsistent deviation direction is marked as pending review. A task snapshot set is formed based on the error-free state set as the state judgment basis for scheduling execution;

[0100] In the later stage of scheduling feedback and strategy updating, the scheme adds the state consistent node to the continue execution queue through real-time comparison and marking, and immediately stops and performs node-by-node state recovery according to the snapshot record as soon as a state inconsistent node is encountered. At the same time, the system records whether the state recovery behavior appears continuously in the scheduling path, and constructs a path stability index to identify high-risk paths. If a path triggers recovery continuously, the system resets the task process chain parameters thereof, and updates the path structure by reconfiguring the target state, error range and recovery priority. For stable paths, the original setting is maintained, but the scheduling priority is adjusted to ensure robustness.

[0101] The scheme solves the problem of inaccurate task path screening in cross-platform data calling by introducing the dependence strength as the evaluation value of the scheduling path; the generation and comparison of the state snapshot before the task is started are realized by setting the target state, error range and recovery priority of each data processing node, and the problem that the state exception is difficult to be found in advance is solved; in the task execution process, the node-by-node recovery of the interrupted path is supported by combining the state check record and the task snapshot set, and the problem of path disaster recovery response lag is solved; finally, the path stability index is constructed through the recovery trigger frequency, and the problems of path reliability that cannot be dynamically judged and evaluation value update lag are solved.

[0102] Finally: the above only for the preferred embodiments of the present application, and not for limiting the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for enterprise data collaborative scheduling for cross-platform integration, characterized in that, include: S1. Obtain task execution records from each business platform, parse the cross-platform data call order, construct a set of scheduling paths composed of data processing nodes, and calculate the dependency strength of each path as an evaluation value. S2. Select paths with evaluation values ​​higher than the threshold from the set of scheduling paths, set the target status, error range and recovery priority of each data processing node, and generate a task process chain according to the data processing node calling order and platform information. S3. Before the task starts, call the status monitoring interface of each business platform to obtain the structured status data of each data processing node in the task process chain for the current period, as a snapshot of the current status, and compare the fields with the target status of the corresponding node to generate a status verification record. S4. Organize the current state snapshot and state verification record into a task snapshot set according to the data processing node call order, which is used for data disaster recovery judgment and state recovery. S5. During task execution, the current status snapshot of each data processing node is compared with the status verification record in the task snapshot set in real time according to the task process chain. If the status matches, the data processing node continues to execute; if the status is abnormal, the current path is terminated, and the status of each node is restored node by node according to the snapshot record. S6. After the task is completed, update the evaluation value of the path in the scheduling path set according to whether the path has undergone a recovery operation; mark the path that has triggered recovery twice in a row as an adjustment object and reset its task process chain parameters; keep the current settings for the path that has not triggered recovery.

2. The enterprise data collaborative scheduling method for cross-platform integration according to claim 1, characterized in that: In S1, by reading the task execution records of each business platform, the data processing nodes involved in each task are obtained, and the data are organized into a task path sequence according to the execution order of the tasks. The data processing nodes contained in each task in the task path sequence are arranged into a node sequence according to the calling order, and adjacent data processing nodes are connected to form a node connection relationship. All node connection pairs are extracted and the output is a set of directed edges. For each node connection pair in the set of directed edges, count its frequency of occurrence in the task path sequence, perform weight calculation based on the frequency, and append the calculation result to the corresponding node connection pair to form a weighted set of directed edges. Connect the traversable data processing nodes in the weighted directed edge set to form node paths. Accumulate the weight values ​​of each edge for each path to obtain the dependency strength value of the path. Form a scheduling path set by combining all paths and their dependency strengths, and use the dependency strength value as the evaluation value of each path.

3. The enterprise data collaborative scheduling method for cross-platform integration according to claim 2, characterized in that: In S2, the distribution of evaluation values ​​for each path is obtained by reading the dependency strength value corresponding to each path in the scheduling path set. Then, the evaluation value interval division operation is performed, high-risk path segments are identified according to the set dynamic threshold, paths with evaluation values ​​higher than the threshold are filtered out, and a candidate path set is output. By parsing the data processing node call order of each path in the candidate path set, the data processing nodes appearing sequentially in the path are identified, and the business platform identifiers corresponding to each node are extracted. The data processing node information with platform identifiers is output in sequence. The data processing node information is input into the target state setting process, and a three-stage configuration operation is performed. The three-stage configuration operation includes setting the target state of the data processing node based on the platform's historical tasks, setting the error tolerance range based on the sensitivity of the data processing node, allocating recovery priorities according to the path interruption history, and outputting a node information sequence containing complete configuration parameters. The node information sequence containing complete configuration parameters is concatenated into a task process chain in path order, where each data processing node includes node identifier, platform identifier, target status, error range, and recovery priority.

4. The enterprise data collaborative scheduling method for cross-platform integration according to claim 3, characterized in that: In S3, by traversing each data processing node in the task process chain, calling the status monitoring interface of the corresponding business platform, collecting the structured status data of each data processing node in the current running cycle, and recording them in the order of the data processing nodes in the task process chain, a current status snapshot is generated. The current state snapshot is compared with the target state sequence set in the task process chain at the field level. The state deviation value and deviation direction of each data processing node are calculated. Before the field-level comparison, the field normalization function is called in advance to uniformly map the structured state fields of cross-platform data processing nodes to solve the differences in data expression between platforms. The deviation value is calculated from the numerical difference between the current value of the normalized field and the target value. The deviation direction is determined according to the sign of the difference. The absolute value of the state deviation value of each data processing node is compared with the corresponding set error range. If the absolute value of the state deviation value is greater than the error range, the data processing node is marked as out of limit, and the abnormality type is marked in combination with the deviation direction. If the state deviation value is within the error range, it is marked as a normal state; If the state data collection fails, it will be marked as an undeterminable state. For nodes marked as undeterminable, a delay determination mechanism will be triggered during the execution of subsequent scheduling paths. They will not be included in the state consistency comparison scope for the time being, so as to ensure the continuity and stability of the overall scheduling chain. According to the order of data processing node calls in the task process chain, the status marking results of each node are combined in sequence to form a status comparison label sequence, and this label sequence, together with the current status snapshot and the target status sequence, is recorded as a status verification record.

5. The enterprise data collaborative scheduling method for cross-platform integration according to claim 4, characterized in that: In S4, by reading the order of each data processing node in the task process chain, the corresponding current state snapshot and state verification record are extracted in sequence, and the two types of state information are paired and combined according to the node call order to form a state comparison sequence. The state comparison sequence is divided into several sliding comparison intervals according to the arrangement order of the data processing nodes. Each interval contains a fixed number of consecutive state pairs. Then, in each comparison interval, the matching consistency between the current state and the state verification mark is compared sequentially, and all mark conflict positions are recorded. For the data processing node corresponding to the identified conflict location, continue to read its target state and error range, and calculate the deviation direction of the current state relative to the target state; if the deviation direction is inconsistent with the deviation direction recorded in the state verification record, then replace the state mark of the node in the state comparison sequence with the state to be reviewed. All status comparison items that are not marked as pending review are recombined according to the node call order in the task process chain to construct a task snapshot set, and the task snapshot set and conflict node index information are output together.

6. The enterprise data collaborative scheduling method for cross-platform integration according to claim 5, characterized in that: In S5, by sequentially traversing each data processing node in the task process chain, the current state snapshot of each data processing node in the current task cycle is compared with the corresponding state verification record in the task snapshot set at the field level, and the judgment flag of each comparison result is recorded in real time during the comparison process. Each data processing node whose field-level comparison results are consistent is added to the continue execution queue of the current scheduling path in the order of the task process chain, and the comparison operation of subsequent nodes is continuously executed. If the status of all nodes is consistent, the task process chain completes the data scheduling according to the original path. If the status verification record of any data processing node is marked as inconsistent during the comparison process, the execution process of the current scheduling path is immediately stopped, the current status snapshot content of each inconsistent node is extracted from the task snapshot set in sequence, and the structured status data in the snapshot is loaded into the corresponding data processing node. After the data processing node recovers its state, the record of the state consistency comparison result of the scheduling path during task execution is updated in real time, and the path in which two or more consecutive data processing nodes recover their states is marked as a high-risk path.

7. The enterprise data collaborative scheduling method for cross-platform integration according to claim 6, characterized in that: In S6, after each scheduling path completes its scheduling task, it traverses each data processing node in the task process chain, extracts its corresponding status verification record, determines whether the scheduling path has a node status recovery operation triggered based on the status deviation value, and uses the recovery trigger result as a risk label for the path scheduling history. Based on the risk labels recorded in the scheduling history, the recovery trigger frequency of each scheduling path in the current and previous scheduling cycles is calculated, and a path scheduling stability index is constructed. If the stability index of the scheduling path meets the judgment condition of continuous triggering recovery, the path is marked as a risk-increasing path, and the target state, error range and recovery priority of the data processing node are reconfigured based on its current task process chain parameters to form an updated task process chain. If the stability index of a scheduling path is lower than the abnormal intervention threshold and it does not trigger any recovery operation in the current period, its evaluation value will remain unchanged and used for subsequent scheduling priority ranking, and the priority will be re-ranked according to the stability index of each path in the current period.

8. An enterprise data collaborative scheduling system for cross-platform integration, comprising a path extraction module, a path filtering module, a status acquisition module, an information integration module, an exception handling module, and a path update module, characterized in that: The path extraction module obtains the task execution records of each business platform, parses the cross-platform data call order, constructs a set of scheduling paths composed of data processing nodes, and calculates the dependency strength of each path as an evaluation value. The path filtering module filters paths with evaluation values ​​higher than the threshold from the set of scheduled paths, sets the target status, error range and recovery priority of each data processing node, and generates a task process chain according to the data processing node calling order and platform information. The status acquisition module obtains the structured status data of each data processing node in the task process chain in the current period by calling the status monitoring interface of each business platform before the task starts. This data is taken as a snapshot of the current status and compared with the target status of the corresponding node to generate a status verification record. The information integration module organizes the current state snapshot and state verification record into a task snapshot set according to the data processing node call order, which is used for data disaster recovery judgment and state recovery. The exception handling module compares the current state snapshot of each data processing node with the state verification record in the task snapshot set in real time according to the task process chain during task execution. If the state matches, the node continues to execute; if the state is abnormal, the current path is terminated and the state of each node is restored node by node according to the snapshot record. The path update module updates the evaluation value of the path in the scheduling path set after the task is completed, depending on whether the path has undergone a recovery operation; it marks the path that has triggered recovery twice in a row as an adjustment object and resets its task process chain parameters; and it keeps the current settings for paths that have not triggered recovery for continuous scheduling.

Images