Transaction system link operation and maintenance method and device based on data stream fusion, and storage medium

Abstract

The invention discloses a transaction system link operation and maintenance method and device based on data stream fusion and a storage medium, and belongs to the technical field of financial data management. Acquiring a multi-source call chain according to the service call information; and according to a preset data stream generation rule and a time delay rule, associating the related call chains of which the end-to-end time delay does not exceed a first time delay threshold value into a data stream in real time, and aggregating the related call chains of which the end-to-end time delay exceeds the first time delay threshold value and the penetration time delay between call nodes does not exceed a second time delay threshold value into the data stream at regular time, the data flow generation rule associates the call chain through a service data mark, and the time delay rule limits the end-to-end time delay of the call chain and the penetration time delay between call nodes; according to the preset alarm rule, abnormity alarm is carried out on the data flow generation event, segmented and flexible link index real-time analysis is provided for a complex transaction link scene, and the reliability of link operation and maintenance is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of financial data management, and in particular relates to a transaction system link operation and maintenance method, device and storage medium based on data stream fusion. Background Art

[0002] Industry APMs primarily use TraceID to track and process RPC call chains, which is very effective in microservice architectures. However, securities trading chains involve not only RPC calls but also asynchronous processing, push, and multicast modes. The data flow in this process is often fragmented by external systems and cannot be connected in series using a single TraceID.

[0003] Data loss or latency in transaction chains requires rapid detection, localization, and resolution. However, scenario and technical analysis reveals the unique nature of transaction frameworks and chains, necessitating observability solutions to address the following pain points to enable rapid problem detection and localization in transaction systems: 1. Inability to track multiple call chains simultaneously: A complete transaction chain is typically not captured in a single trace, requiring aggregation of multiple traces based on business context. 2. Insufficient granularity in tracking metrics: Transactions require attention to metrics spanning end-to-end latency and single-item penetration latency, both of which are equally important. Traditional request-response-based span designs cannot fully meet these requirements. 3. Limited troubleshooting methods for call issues: Troubleshooting call issues across transactions and microservice chains requires inefficient methods like logs due to data disconnection between transaction and microservice chains. 4. Lack of transaction chain analysis capabilities: Troubleshooting transaction chain latency requires layer-by-layer analysis, from the client side, gateway, message center, and order system. Furthermore, this analysis is limited to single transactions and lacks statistical analysis capabilities. Summary of the Invention

[0004] The purpose of the present invention is to provide a transaction system link operation and maintenance method, device and storage medium based on data stream fusion, which associates data streams from different sources through business data tags, thereby providing a complete transaction link view, and provides segmented and flexible link indicator real-time observation and analysis for complex transaction link scenarios based on custom configuration rules.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] In a first aspect, the present invention provides a transaction system link operation and maintenance method based on data stream fusion, comprising: Obtain business call information in the transaction system; Obtain a multi-source call chain according to the business call information; Based on pre-established data flow generation rules and delay rules, relevant call chains whose end-to-end delay does not exceed a first delay threshold are associated in real time to form a data flow, and relevant call chains whose end-to-end delay exceeds the first delay threshold and whose penetration delay between call nodes does not exceed a second delay threshold are periodically aggregated to form a data flow. The data flow generation rules associate call chains using service data tags, and the delay rules are used to limit the end-to-end delay of each call chain in the data flow and the penetration delay between call nodes in the call chain. Generate abnormal alarms for data flow events based on pre-built alarm rules.

[0007] Optionally, obtaining the service call information in the transaction system includes: The business call information reported by the synchronous call scenario is collected through the Trace SDK, and the business call information reported by the asynchronous call scenario is collected according to the pre-built asynchronous data reporting mechanism. The business call log is generated based on the collected business call information of each call scenario. The asynchronous data reporting mechanism includes a mechanism for determining the timing of asynchronous request data reporting, the calling node's self-confirmation of the call chain ID, and the determination of the call chain call order and nesting level.

[0008] Optionally, obtaining a multi-source call chain according to the business call information includes: Collect link tracing context from business call logs; A call chain is formed based on the call information of related nodes in the link tracking context, wherein the call information of related nodes includes the behavior data of each call node, the call order of each call node, and the nesting level. Optionally, the pre-built data flow generation rules include: Filtering the multiple source call chains according to the topological structure of the call chains, the nodes passed by the call chains, and the additional service tags in the call chains to obtain a target call chain; The relevant target call chains are associated into a data stream according to preset call chain association parameters, wherein the association parameters include an additional service tag in the call chain for identifying the data stream.

[0009] Optionally, according to a pre-established data flow generation rule, the step of associating relevant call chains whose end-to-end delay does not exceed a first delay threshold in real time to form a data flow includes: Associating multiple related call chains using a session window to form a data stream, wherein the session window sets a session interval according to a first latency threshold; Write the associated data stream into the ES database in real time.

[0010] Optionally, after correlating relevant call chains whose end-to-end delay does not exceed the first delay threshold in real time to form a data stream, the method further includes: Perform real-time correlation matching on related call chains that are not aggregated into data streams and generate data stream IDs; The data stream ID and the corresponding call chain ID of each call chain are written into the cache database in real time.

[0011] Optionally, the step of periodically aggregating relevant call chains whose end-to-end delay exceeds a first delay threshold and whose penetration delay between call nodes does not exceed a second delay threshold to form a data flow according to pre-established data flow generation rules and delay rules includes: Read the data flow ID and its corresponding call chain ID in the cache database; Query the penetration delay between call nodes based on the call chain ID; Periodically aggregate call chains whose penetration delays between call nodes do not exceed the second delay threshold to form a data stream; Write the periodically aggregated data stream into the ES database and delete the corresponding data stream ID and call chain ID in the cache database.

[0012] Optionally, generating an abnormal alarm for a data flow generation event according to a pre-established alarm rule includes: Writing the second delay threshold as an alarm indicator into the alarm data channel; An alarm is generated for a data flow whose penetration delay between calling nodes exceeds a second delay threshold.

[0013] In a second aspect, the present invention provides a transaction system link operation and maintenance device based on data stream fusion, comprising: Call information acquisition module: used to obtain business call information in the transaction system; Call chain acquisition module: used to obtain multi-source call chains based on the business call information; Data flow aggregation module: used to associate relevant call chains whose end-to-end delay does not exceed the first delay threshold in real time to form a data flow based on pre-established data flow generation rules and delay rules, and periodically aggregate relevant call chains whose end-to-end delay exceeds the first delay threshold and whose penetration delay between call nodes does not exceed the second delay threshold to form a data flow, wherein the data flow generation rules associate call chains by business data tags, and the delay rules are used to limit the end-to-end delay of each call chain in the data flow and the penetration delay between call nodes in the call chain; Abnormal alarm module: used to generate abnormal alarms for data flow events based on pre-built alarm rules.

[0014] In a third aspect, the present invention provides a computer storage medium having a computer program stored thereon. When the computer program is executed by a processor, the transaction system link operation and maintenance method based on data stream fusion as described in any one of the first aspects is implemented.

[0015] Compared with the existing technology, the beneficial effects achieved by the present invention are as follows: by introducing user-defined data flow configuration items and delay configuration items into the process of generating data flow by the call chain, the delay problem can be located for the call chain in both synchronous call scenarios and asynchronous call scenarios, and then combined with the alarm configuration items to perform full-link alarms for data flow, thereby reducing the time and threshold for troubleshooting full-link delay problems of data flow, and settling the original troubleshooting process that relied on people and experience to a tool platform. Through the platform, link delay problems can be quickly located and troubleshooted, thereby improving the efficiency of problem discovery and the control over the system operation status. Link bottlenecks can be identified through data statistical analysis capabilities, guiding the optimization and improvement of the entire system link. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 FIG2 is a flow chart of a transaction system link operation and maintenance method based on data stream fusion in Example 1 of the present invention;

[0017] Figure 2 FIG2 is a flow chart of a transaction system link operation and maintenance method based on data stream fusion in Example 2 of the present invention;

[0018] Figure 3 The figure shows an architecture diagram of calling the information collection terminal Trace SDK in one embodiment of the present invention;

[0019] Figure 4 The figure shows the chainId generation process for asynchronous request response and asynchronous push scenarios in one embodiment of the present invention;

[0020] Figure 5 The figure shows the chainId generation process in a multicast push scenario in one embodiment of the present invention;

[0021] Figure 6 Shown is a flow chart of data stream processing performed by a data stream engine in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention.

[0023] Example 1

[0024] like Figure 1 As shown, this embodiment provides a transaction system link operation and maintenance method based on data stream fusion, including: Obtain business call information in the transaction system; Obtain a multi-source call chain according to the business call information; Based on pre-established data flow generation rules and delay rules, relevant call chains whose end-to-end delay does not exceed a first delay threshold are associated in real time to form a data flow, and relevant call chains whose end-to-end delay exceeds the first delay threshold and whose penetration delay between call nodes does not exceed a second delay threshold are periodically aggregated to form a data flow. The data flow generation rules associate call chains using service data tags, and the delay rules are used to limit the end-to-end delay of each call chain in the data flow and the penetration delay between call nodes in the call chain. Generate abnormal alarms for data flow events based on pre-built alarm rules.

[0025] By linking data streams from different sources through business data tags, a complete view of the transaction chain is provided. In addition to data association, the system provides segmented and flexible real-time observation and analysis capabilities for chain indicators in complex transaction chain scenarios. It supports real-time statistical analysis of these multiple data streams, transforming complex chain issues that originally required multi-person collaboration into platform-based, automated processing, helping users identify and resolve potential issues more quickly.

[0026] Example 1

[0027] Based on Example 1, this example also makes the following design.

[0028] Based on the fusion association of multiple data sources and Figure 2 The directed graph shown uses the multi-source call chain and business request information reported in the business process to trigger the generation of session data for each graph node according to the static flowchart, dynamically track each session, and ultimately achieve dynamic observability of the business process.

[0029] like Figure 3 As shown, this embodiment uses the Trace SDK to collect call information. The Trace SDK uses a layered architecture and relies on the trace core module at the bottom layer. This module encapsulates the underlying mechanisms such as the generation of link context information (traceId, chainId) and data reporting.

[0030] Tracer is a simple, generalized interface that builds spans, transmits spans, and samples and reports spans. As shown in Table 1, the corresponding Tracer implementation class is implemented according to different call scenarios, encapsulating the chainId generation rules. These include: synchronous call SyncTracer; asynchronous call AsyncTracer; PushTracer; and MulticastTracer.

[0031] Table 1 method describe TraceSpan clientSend() The client sends a request span (synchronous client request, asynchronous client request, asynchronous client push, multicast client push) void clientReceive(TraceSpantraceSpan) The client receives the response span (involved in synchronous client receiving response and asynchronous client receiving response scenarios) TraceSpan serverReceive(TraceSpanCore traceSpanCore) Server receiving request span (synchronous server receiving request, asynchronous server receiving request, asynchronous server receiving push, multicast server receiving multicast) void serverSend(TraceSpan traceSpan) The server sends a response span (synchronous server sends a response, asynchronous server sends a response)

[0032] TraceSpan is a span unit. In actual application, a span is a complete data packet that contains the complete data that the current node needs to report. Table 2 shows the code corresponding to obtaining span information and adding additional tags.

[0033] Table 2 method describe TraceInfo getTraceInfo() Get the complete information of span, the corresponding object is TraceInfo TraceSpanCore getSpanCore() Get the span transmission information, the corresponding object is TraceSpanCore addTraceInfoTags(Map<String, Object>map) Add a tag

[0034] TraceContext is crucial for implementing tracebacks. It enables sharing of thread context traceback data. Table 3 shows the code for reading and writing span information in the context.

[0035] Table 3 method describe TraceSpan getSpan() Get span information in context setSpan(TraceSpan span) Write span information to the context

[0036] In this embodiment, the APM collection layer has built an SDK mainly for self-developed Java microservices, and the communication protocols are mainly Dubbo and HTTP. However, trading systems that mainly rely on asynchronous calls cannot directly reuse the existing SDK capabilities, and the collected data of such trading systems needs to be adapted. The following is the data reporting mechanism in asynchronous scenarios:

[0037] (1) Timing of asynchronous request data reporting

[0038] Asynchronous request-response scenario: The client sends a request (cs), the client receives a response (cr), the server receives the request (pr), and the server sends a response (ps).

[0039] Asynchronous push scenario: The server sends push (ps) and the client receives push (cr).

[0040] (2) The traceId is the unique identifier of the call chain. When each node sends a request or message, it decides whether to generate a new traceId or reuse the upstream traceId.

[0041] (3) chainId is used to distinguish the calling order and nesting level in the calling chain. Figure 4As shown in the figure, the chainID generation rule in the asynchronous scenario is: in the request and push phases, the chainID starts at 0, and the chainID of the subsequent links is: [parent link chainID] + "." + [child link call order]. In the response phase, the chainID starts from the last chainID in the request phase, and the chainID of the subsequent links is: [parent link chainID] - "." - [last call order].

[0042] (4) If Figure 5 As shown in the figure, for multicast push messages, the message receiver rewrites the chainId, rewriting the link ID to a unique identifier string for the multicast receiver (the string cannot contain dots). The receiving end can be distinguished based on this string. The APM receiving end performs the merging process.

[0043] (5) The data reporting protocol adds a message type field. This field distinguishes between synchronous requests and asynchronous requests, asynchronous responses, asynchronous push, and asynchronous multicast. To be compatible with the current synchronous version, this field is not required and the default is synchronous request.

[0044] Each link can be assigned one or more tags. In transaction scenarios, business fields such as orderId and tradingAccount can be used to associate different "chains" to facilitate query and analysis.

[0045] like Figure 6 As shown, in this embodiment, a data flow engine is used to aggregate multiple call chains into a data flow according to user configuration items, and a real-time alarm engine is used to generate delay alarms and data flow integrity alarms. The specific steps are as follows:

[0046] 1. Flink real-time tasks use the broadcast stream mechanism to periodically pull and update user configuration items and push them to processing operators in the form of broadcasts. User configuration items include data stream generation rules, custom delay rules, and alarm rules, which are parsed and saved in the MySQL database.

[0047] Data flow generation rules, which determine how to filter out call chains and associate multiple call chains to form a single data flow, can be divided into filtering rules and association rules. Filtering rules filter target call chains based on their topology, the IP addresses they pass through, and the tags in the call chains. Association rules associate multiple call chains to form a data flow based on user-specified association parameters. Association parameters, defined as custom business tags attached to the call chains, are typically unique business IDs like order numbers. They also uniquely identify the data flow.

[0048] The custom delay rule describes the start and end points of the delay within this data flow.

[0049] For defined custom latency, you can configure indicator alarm rules, such as indicator threshold alarms.

[0050] 2. Flink real-time tasks associated with data streams: consume call chain data in real time, use session windows for link pre-aggregation, and write to the ES database in real time.

[0051] The Flink real-time task associated with the data stream periodically pulls the rules created by the user and saves the ID of the associated data stream and the corresponding call chain ID to Redis.

[0052] 3. The Flink task matches the link data that needs to generate data streams in real time, periodically pulls the rules created by the user, generates a data stream ID based on the business field, and writes the mapping relationship between the data stream ID and multiple TraceIds into the Redis cache database in real time.

[0053] The core data stored in Redis is: (1) Data Stream ID: the unique identifier of the data stream with the most recently collected call chain; (2) Call Chain ID: the call chain ID of each data stream that has been recently collected. One of the core capabilities of the data stream engine is to aggregate multiple asynchronous call chains with long time intervals into a single data stream.

[0054] 4. The introduction of the Redis cache database is to improve the aggregation efficiency of data streams and to hot cache data that has not been aggregated into data streams. With the help of Redis, multiple call chain IDs can be collected and associated into a data stream.

[0055] 5. Flink asynchronous scheduled task for aggregated data streams: This task periodically reads and queries the unaggregated data stream data in Redis. For data streams that have not been aggregated or have exceeded the user-specified timeout, the task retrieves the TraceId list from the corresponding list, periodically aggregates the generated data stream data, writes it to ES, and deletes the corresponding Redis cache data.

[0056] Asynchronous scheduled tasks are divided into a distribution module and a calculation module. The distribution module is responsible for reading the most recently collected data stream information from Redis and distributing this information to the calculation module via Kafka. The calculation module is responsible for aggregating the data stream. The calculation module of asynchronous scheduled tasks uses the data stream ID and call chain ID consumed from Kafka to query the database for call chain details. It then aggregates the data stream and calculates the customized penetration delay based on data stream rules and customized delay rules.

[0057] 6. Combine the custom penetration delay configuration with the generated data stream to generate indicators, write the delay indicator data into the indicator alarm and indicator storage data channel, complete the indicator storage and indicator alarm, reuse the indicator writing and indicator alarm capabilities, and store and alarm the custom penetration delay data.

[0058] 7. Generate events for data flows on links that do not conform to user configurations and provide real-time alarm notifications.

[0059] In this embodiment, a visual interface is also provided to display the analysis results in an intuitive manner, thereby improving the efficiency of analysis and processing.

[0060] This embodiment utilizes data storage and real-time big data analysis and alerting capabilities to enhance the digital operation and maintenance monitoring and observability of the trading system. This enables comprehensive real-time statistical analysis of the system's operating status, integrated real-time monitoring, and rapid troubleshooting, significantly improving the trading system's operational risk identification capabilities and emergency response efficiency. In terms of basic capabilities, it supports the orchestration and correlation analysis of multiple data streams. For problem detection, it can monitor various indicators of multiple data stream combinations in real time, including traffic, latency, and error rates, to promptly identify anomalies. Furthermore, it provides a flexible alerting mechanism that quickly triggers alerts when potential problems are detected, notifying relevant personnel to address them. For problem location, it can record and display the complete call chain of transaction requests in the chain, including key information such as the call relationship and time consumption of each service node. This allows for rapid location of specific service nodes when problems arise, narrowing the scope of investigation. For problem analysis, it provides powerful data analysis capabilities that can deeply mine and analyze collected data, helping users identify the root causes and bottlenecks of problems.

[0061] Example 3

[0062] This embodiment provides a transaction system link operation and maintenance device based on data stream fusion, including: Call information acquisition module: used to obtain business call information in the transaction system; Call chain acquisition module: used to obtain multi-source call chains based on the business call information; Data flow aggregation module: used to associate relevant call chains whose end-to-end delay does not exceed the first delay threshold in real time to form a data flow based on pre-established data flow generation rules and delay rules, and periodically aggregate relevant call chains whose end-to-end delay exceeds the first delay threshold and whose penetration delay between call nodes does not exceed the second delay threshold to form a data flow, wherein the data flow generation rules associate call chains by business data tags, and the delay rules are used to limit the end-to-end delay of each call chain in the data flow and the penetration delay between call nodes in the call chain; Abnormal alarm module: used to generate abnormal alarms for data flow events based on pre-built alarm rules.

[0063] Example 4

[0064] This embodiment provides a computer storage medium having a computer program stored thereon. When the computer program is executed by a processor, the transaction system link operation and maintenance method based on data stream fusion as described in any step of Example 2 is implemented.

[0065] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0066] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0067] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0068] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0069] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which are all protected by the present invention.

Claims

1. A transaction system link operation and maintenance method based on data stream fusion, characterized in that: include: Obtain business call information in the transaction system; Obtain a multi-source call chain according to the business call information; Based on pre-established data flow generation rules and delay rules, relevant call chains whose end-to-end delay does not exceed a first delay threshold are associated in real time to form a data flow, and relevant call chains whose end-to-end delay exceeds the first delay threshold and whose penetration delay between call nodes does not exceed a second delay threshold are periodically aggregated to form a data flow. The data flow generation rules associate call chains using service data tags, and the delay rules are used to limit the end-to-end delay of each call chain in the data flow and the penetration delay between call nodes in the call chain. Generate abnormal alarms for data flow events based on pre-built alarm rules.

2. The transaction system link operation and maintenance method based on data stream fusion according to claim 1 is characterized in that: The obtaining of the business call information in the transaction system includes: The business call information reported by the synchronous call scenario is collected through the Trace SDK, and the business call information reported by the asynchronous call scenario is collected according to the pre-built asynchronous data reporting mechanism. The business call log is generated based on the collected business call information of each call scenario. The asynchronous data reporting mechanism includes a mechanism for determining the timing of asynchronous request data reporting, the calling node's self-confirmation of the call chain ID, and the determination of the call chain call order and nesting level.

3. The transaction system link operation and maintenance method based on data stream fusion according to claim 2 is characterized in that: The acquiring of a multi-source call chain according to the business call information includes: Collect link tracing context from business call logs; A call chain is formed according to the relevant node call information in the link tracking context, wherein the relevant node call information includes the behavior data of each call node, the call order of each call node, and the nesting level.

4. The transaction system link operation and maintenance method based on data stream fusion according to claim 1 is characterized in that: The generating rules based on the pre-built data stream include: Filtering the multiple source call chains according to the topological structure of the call chains, the nodes passed by the call chains, and the additional service tags in the call chains to obtain a target call chain; The relevant target call chains are associated into a data stream according to preset call chain association parameters, wherein the association parameters include an additional service tag in the call chain for identifying the data stream.

5. The transaction system link operation and maintenance method based on data stream fusion according to claim 4 is characterized in that: The steps of correlating relevant call chains whose end-to-end delay does not exceed a first delay threshold in real time to form a data flow according to a pre-established data flow generation rule include: Associating multiple related call chains using a session window to form a data stream, wherein the session window sets a session interval according to a first latency threshold; Write the associated data stream into the ES database in real time.

6. The transaction system link operation and maintenance method based on data stream fusion according to claim 5 is characterized in that: After the relevant call chains whose end-to-end delay does not exceed the first delay threshold are correlated in real time to form a data stream, the method further includes: Perform real-time correlation matching on related call chains that are not aggregated into data streams and generate data stream IDs; The data stream ID and the corresponding call chain ID of each call chain are written into the cache database in real time.

7. The transaction system link operation and maintenance method based on data stream fusion according to claim 6 is characterized in that: The steps of periodically aggregating relevant call chains whose end-to-end delay exceeds a first delay threshold and whose inter-call node penetration delay does not exceed a second delay threshold to form a data flow according to pre-established data flow generation rules and delay rules include: Read the data flow ID and its corresponding call chain ID in the cache database; Query the penetration delay between call nodes based on the call chain ID; Periodically aggregate call chains whose penetration delays between call nodes do not exceed the second delay threshold to form a data stream; Write the periodically aggregated data stream into the ES database and delete the corresponding data stream ID and call chain ID in the cache database.

8. The transaction system link operation and maintenance method based on data stream fusion according to claim 1 is characterized in that: The abnormal alarm generated by the data flow event according to the pre-built alarm rules includes: Writing the second delay threshold as an alarm indicator into the alarm data channel; An alarm is generated for a data flow whose penetration delay between calling nodes exceeds a second delay threshold.

9. A transaction system link operation and maintenance device based on data stream fusion, characterized in that: include: Call information acquisition module: used to obtain business call information in the transaction system; Call chain acquisition module: used to obtain multi-source call chains based on the business call information; Data flow aggregation module: used to associate relevant call chains whose end-to-end delay does not exceed the first delay threshold in real time to form a data flow based on pre-established data flow generation rules and delay rules, and periodically aggregate relevant call chains whose end-to-end delay exceeds the first delay threshold and whose penetration delay between call nodes does not exceed the second delay threshold to form a data flow, wherein the data flow generation rules associate call chains by business data tags, and the delay rules are used to limit the end-to-end delay of each call chain in the data flow and the penetration delay between call nodes in the call chain; Abnormal alarm module: used to generate abnormal alarms for data flow events based on pre-built alarm rules.

10. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the transaction system link operation and maintenance method based on data stream fusion according to any one of claims 1 to 8 is implemented.

Images