A method and device for synchronizing main and backup data of a securities transaction system

Abstract

This invention discloses a method and apparatus for master-slave data synchronization in a securities trading system. The method first involves the master node receiving external input messages, then generating time messages at fixed time intervals on its own machine. The external input messages and time messages are then ordered to form a sequenced message sequence. The master node sends this sequenced message sequence to the slave node, simultaneously processing its own business logic based on the message sequence. The slave node receives the message sequence, processes it according to the same business logic as the master node, and restores its memory data. Finally, the slave node processes the message sequence. This invention decouples the master-slave data synchronization mechanism from the business logic, avoiding the impact of business changes on the synchronization mechanism, while ensuring strict consistency of the calculation results between the master and slave nodes, thus improving the stability and accuracy of the securities trading system's high availability.

Description

Technical Field

[0001] This invention relates to the field of securities trading technology, and in particular to a method and apparatus for synchronizing primary and backup data in a securities trading system. It is applicable to the hot standby solution of a brokerage's memory trading system and can also be adapted to the primary and backup data synchronization requirements of a financial trading risk control system. Background Technology

[0002] In the securities trading field, the high availability of brokerage firms' in-memory trading systems relies on a reliable master-slave data synchronization mechanism. In existing technologies, the master node typically synchronizes processed in-memory data, including orders, funds, positions, and intra-node risk control status data, to the slave node. Specifically, after processing risk control orders within its computing node, the master node synchronizes the updated risk control status data to the slave node; after processing orders, it synchronizes the updated order status to the slave node; and after processing funds and positions, it synchronizes the updated funds and positions data to the slave node.

[0003] The existing master-slave data synchronization mechanism has obvious technical defects: the data synchronization logic is coupled in different places in the business processing code, the number of synchronization points is large, and the system complexity is high; when the business logic changes, such as adding risk control rules in the transaction node, the master-slave data synchronization mechanism needs to be modified accordingly, and the backup node also needs to adjust the data update logic accordingly, which greatly increases the complexity of the system's high availability implementation, and the high availability reliability will be affected by the frequent modification of the business logic.

[0004] Meanwhile, securities trading systems involve operations that are strongly correlated with local machine time, such as risk control calculations within trading nodes and time-based order placement and flow control. Even if the primary and backup servers control the time difference to the microsecond level through time synchronization protocols, it is still impossible to guarantee that the calculation results of the primary and backup nodes based on their local time are strictly consistent, which further reduces the accuracy of primary and backup data synchronization. Summary of the Invention

[0005] This invention addresses the shortcomings of existing technologies by providing a method and apparatus for synchronizing primary and backup data in a securities trading system. This method decouples the primary and backup data synchronization mechanism from business logic, avoids the impact of business changes on the synchronization mechanism, and ensures strict consistency of calculation results between primary and backup nodes, thereby improving the high availability, stability, and accuracy of the securities trading system.

[0006] Technical solution: The present invention provides a method for synchronizing primary and backup data in a securities trading system, comprising the following steps:

[0007] Step 1: Receive external input messages through the transaction node master node. The external input messages include order placement, order cancellation, transaction confirmation, and risk control system judgment results.

[0008] Step 2: Generate time messages at fixed time intervals on the local machine through the master node, and sort the external input messages and time messages in a unified order to form a sorted message sequence;

[0009] Step 3: The master node sends the ordered message sequence to the transaction node backup node, and the master node completes its own business logic processing based on the message sequence.

[0010] Step 4: Receive the message sequence through the backup node, process the message sequence according to the same business processing logic as the master node, restore the memory data of the backup node, and the backup node does not send messages to external systems during the processing.

[0011] Step 5: When the backup node is processing the message sequence, all time-triggered judgment operations are executed based on the time message in the message sequence, rather than the local time of the backup node.

[0012] Furthermore, the generation interval of the time message in step 2 can be adjusted according to actual business needs. The reception of external input messages and the generation of time messages have no fixed order. The sequencing process sorts the two types of messages in a unified manner.

[0013] Furthermore, the method by which the master node sends message sequences to the backup node in step 3 includes asynchronous sending and synchronous sending. The asynchronous sending method is used in low-latency service scenarios.

[0014] Furthermore, the message transmission between the master node and the backup node in step 3 is implemented based on a protocol that supports message persistence, and the protocol layer completes message resending in scenarios of disconnection and message loss.

[0015] Furthermore, the process of restoring the memory data of the backup node in step 4 includes: performing risk control, fund, position and order processing within the node on the entrustment information; performing fund, position and order processing on the risk control judgment results; and performing risk control, fund, position and order processing within the node on the transaction report.

[0016] Furthermore, in step 5, the backup node updates the relative time of the memory record according to the received time message, and performs memory calculations for risk control rule judgment and rate limiting judgment based on the relative time.

[0017] Furthermore, the method also includes an initialization step to ensure that the initialization state of the backup node is completely consistent with that of the master node.

[0018] Furthermore, the method is adapted to the master-slave data synchronization of the financial transaction risk control system. The master node of the risk control system receives external messages and sends them to the backup node of the risk control system after sequencing the local time message. The backup node restores the memory data according to the same logic.

[0019] This invention also discloses a device for synchronizing primary and backup data in a securities trading system, comprising a primary node module and a backup node module, wherein the primary node module and the backup node module are communicatively connected; the primary node module includes: a message receiving unit for receiving external input messages such as order placement, order cancellation, transaction reports, and risk control system judgment results; a time message generation unit for generating time messages at fixed time intervals, the generation interval of which can be adjusted as needed; a sequencing unit for uniformly sequencing the external input messages and time messages to form a message sequence; and a business processing unit for completing the primary node's own business logic processing based on the message sequence and updating the primary node's memory. The system includes: a message sending unit for sending ordered message sequences to the backup node module, supporting both asynchronous and synchronous sending modes; a message receiving unit for receiving ordered message sequences sent by the master node module; a time parsing unit for extracting time messages from the message sequences to provide logical time basis for time trigger judgment; a business processing unit for processing the message sequences according to the same business processing logic as the master node module, only restoring the backup node's memory data and not sending messages to external systems; and a memory recovery unit for updating the backup node's memory data on funds, positions, orders, and risk control status according to the business processing results.

[0020] Furthermore, the communication link between the master node module and the backup node module is equipped with a transmission protocol that supports message persistence. The protocol layer has the functions of reconnection after disconnection and resending lost messages. The device can be adapted and deployed in securities trading systems and financial trading risk control systems.

[0021] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0022] 1. This invention achieves decoupling of the synchronization mechanism from business logic: This invention synchronizes the ordered message sequence from the master node to the backup node, and the backup node autonomously recovers the memory data according to the same logic, which replaces the method of synchronizing the memory data in the prior art. This makes the master-backup data synchronization mechanism independent of the business logic. Changes in business logic do not require modification of the synchronization mechanism, reducing the complexity of system development and maintenance.

[0023] 2. Reduce workload for business iteration: When iterating business requirements and modifying rules, business logic developers do not need to pay attention to the synchronization of primary and backup data. They only need to ensure that the business processing logic of the primary and backup nodes is consistent, which greatly reduces the development and testing workload of business iteration.

[0024] 3. Improve system stability: The master-slave data synchronization mechanism remains stable as a function of the system framework layer and will not be modified with frequent changes in business logic, effectively improving the high availability, stability and reliability of the securities trading system during business evolution.

[0025] 4. Ensure strict consistency of calculation results between primary and backup nodes: By inserting the time message of the primary node into the message sequence, the backup node completes all time-triggered judgments and calculations based on the time message, which completely solves the problem of inconsistent calculation results caused by slight differences in the local time of the primary and backup nodes, and ensures strict synchronization of memory data between the primary and backup nodes.

[0026] 5. Strong adaptability: The method and apparatus of the present invention are not only applicable to securities firms' in-memory trading systems, but can also be directly adapted to the master-slave data synchronization requirements of related financial systems such as financial trading risk control systems. Deployment can be completed with only minor adjustments, and it has a wide range of application scenarios. Attached Figure Description

[0027] Figure 1 This is a flowchart of the present invention.

[0028] Figure 2 This is a schematic diagram of the overall architecture of the master-slave data synchronization mechanism of the securities trading system of the present invention. Detailed Implementation

[0029] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0030] like Figure 1 As shown, a method for synchronizing primary and backup data in a securities trading system includes the following steps:

[0031] S1. External message reception: The main node of the trading node receives external input messages, which include key messages in the entire securities trading process such as order placement and cancellation orders sent by the upstream system, risk control judgment results returned by the risk control system, and transaction reports sent by the downstream system.

[0032] S2. Time Message Generation and Message Sequencing: The master node generates time messages at fixed time intervals based on its local time. The generation interval of the time messages can be flexibly adjusted according to the actual business requirements for time granularity, such as 250 milliseconds / time. The master node performs unified sequencing on the received external input messages and the generated local time messages. There is no fixed order between the receipt of external input messages and the generation of time messages. The sequencing process performs global sorting on the two types of messages to form a unified ordered message sequence.

[0033] S3. Master Node Processing and Message Sending: The master node completes its own business logic processing based on the ordered message sequence, updating in-memory data such as orders, funds, positions, and risk control status. Simultaneously, the master node sends this ordered message sequence to the backup trading node. The sending method can be asynchronous or synchronous depending on business latency requirements; asynchronous sending is preferred for low-latency business scenarios. Message transmission between the master and backup nodes is implemented based on a protocol that supports message persistence. The protocol layer handles message resending in abnormal scenarios such as connection loss and message loss, ensuring the reliability of message transmission.

[0034] S4. Backup Node Message Reception and Business Processing: The backup node receives the ordered message sequence sent by the master node and processes the message sequence one by one according to the same business processing logic as the master node. During the processing, it only completes the recovery and update of its own memory data and does not send messages to any external systems such as upstream systems, risk control systems, and downstream systems, so as to avoid interfering with the overall securities trading process.

[0035] S5. Consistency processing based on time messages: When the backup node processes the message sequence, all time-triggered judgment operations, such as risk control rule judgment and order rate limiting judgment, are executed based on the time messages in the message sequence, rather than the backup node's local time. The backup node updates the relative time recorded in memory according to the received time messages, and completes all time-related memory calculations based on this relative time, ensuring strict consistency of the calculation results between the primary and backup nodes.

[0036] Furthermore, during the opening phase of the securities trading system, the opening initialization operation of the backup node is completed, ensuring that the opening initialization state of the backup node is completely consistent with that of the master node, thus laying the foundation for subsequent master-backup data synchronization.

[0037] Furthermore, this method can be directly adapted to the master-slave data synchronization of financial transaction risk control systems. The master node of the risk control system receives external messages such as order messages and transaction execution report messages sent by multiple transaction systems, generates time messages with fixed time intervals on its own machine, and sequences them uniformly with the external messages. The sequenced message sequence is then sent to the backup node of the risk control system. The backup node processes the message sequence according to the same calculation logic as the master node, restores the memory data, and realizes the master-slave data synchronization of the risk control system.

[0038] Figure 2 This is a schematic diagram of the overall architecture of the primary and backup data synchronization mechanism of the securities trading system of the present invention. The diagram includes five core modules: upstream system, risk control system, downstream system, trading node (primary), and trading node (backup). The interaction relationships and message flow of each module are as follows:

[0039] The upstream system sends order (submission, cancellation) messages to the transaction node (master);

[0040] The transaction node (master) sends a transaction order message to the risk control system, and the risk control system returns a risk control result message to the transaction node (master).

[0041] The trading node (master) sends an order message to the downstream system, and the downstream system returns a transaction confirmation message to the trading node (master).

[0042] The transaction node (master) performs order processing on the received external messages such as orders, risk control results, and returns, and the time messages generated on its own machine.

[0043] The primary transaction node sends the ordered message sequence (including external messages and time messages) to the secondary transaction node.

[0044] The transaction node (backup) receives the message sequence and completes the memory data recovery, without sending messages to any external system.

[0045] The present invention discloses a device for master-slave data synchronization in a securities trading system. The device includes a master node module and a slave node module. The master node module and the slave node module are connected through a communication link. The communication link carries a transmission protocol that supports message persistence. The protocol layer has the functions of reconnection after disconnection and resending lost messages.

[0046] Master node module

[0047] Message receiving unit: Used to receive external input messages throughout the entire securities trading process, including order placement, order cancellation, transaction confirmation, and risk control system judgment results, to achieve unified access to external messages.

[0048] Time Message Generation Unit: Used to generate time messages at fixed time intervals on the local machine. The generation interval can be flexibly adjusted according to the business's requirements for time granularity to meet the time-triggered calculation needs of different scenarios.

[0049] Sequencing Unit: Used to uniformly sequence the external input messages received by the message receiving unit and the time messages generated by the time message generation unit to form a global ordered message sequence, ensuring the timing consistency of message processing.

[0050] Business processing unit: Based on the message sequence generated by the sequencing unit, it completes the securities trading business logic processing of the master node itself, including risk control calculation within the node, fund / position update, order status maintenance, etc., and updates the master node's memory data.

[0051] Message sending unit: Used to send the ordered message sequence to the backup node module. It supports both asynchronous and synchronous sending modes and can be flexibly switched according to the latency requirements of the business. Asynchronous sending is used in low-latency scenarios.

[0052] Backup node module

[0053] Message receiving unit: Used to receive the ordered message sequence sent by the message sending unit of the master node module to realize message synchronization between master and backup nodes.

[0054] Time parsing unit: Used to extract time messages from the message sequence, providing a unified logical time basis for the time-triggered judgment operation of the backup node, replacing the local time of the backup node.

[0055] Business processing unit: Used to process the received message sequence one by one according to the same business processing logic as the business processing unit of the master node module, so as to ensure that the business processing process of the master and backup nodes is consistent.

[0056] Memory recovery unit: Used to update the memory data of the backup node, such as funds, positions, orders, and risk control status, based on the processing results of the business processing unit, to complete the memory data recovery of the backup node and ensure that the memory status of the primary and backup nodes is consistent.

[0057] This device can be directly deployed in securities trading systems, and it can also be adapted to be deployed in financial trading risk control systems. Only the business processing units of the primary and backup node modules need to be adjusted according to the business logic of the risk control system to achieve data synchronization between the primary and backup nodes.

[0058] Example 1: Implementation of a method for synchronizing primary and backup data in a securities trading system

[0059] Opening Initialization: Before the opening of the A-share trading day, the trading node backup node is initialized to ensure that the backup node's initial status, such as funds, positions, and basic risk control rules, is completely consistent with the main node.

[0060] External message reception: The message receiving unit of the main node of the trading node receives in real time order placement and cancellation messages sent by upstream clients, risk control judgment results messages returned by the risk control system, and transaction feedback messages returned by the exchange's order gateway (downstream system).

[0061] Time message generation and sequencing: The time message generation unit of the master node generates local time messages at 250 millisecond intervals. The sequencing unit sequences the external input messages received in real time and the generated time messages. For example, at time t1, a report entrustment message is received, a time message is generated at t1+100 milliseconds, and a risk control judgment result message is received at t1+200 milliseconds. After sequencing, a message sequence of [report entrustment, time message, risk control judgment result] is formed.

[0062] Master Node Processing and Message Sending: The business processing unit of the master node completes business processing based on the ordered message sequence, including risk control calculation within the node for order submission, fund freezing, order status creation, updating the order risk control status according to the risk control judgment result, and updating the master node's memory data at the same time; the message sending unit of the master node sends the ordered message sequence to the backup node asynchronously via a custom TCP protocol. The protocol layer persists the messages, and if a network disconnection occurs, the untransmitted messages are automatically resent after reconnection.

[0063] Backup Node Processing and Memory Recovery: The backup node's message receiving unit receives the message sequence sent by the master node, and the time parsing unit extracts the time message from the sequence and updates the relative time in the backup node's memory. The backup node's business processing unit processes the message sequence according to the same logic as the master node, performs the same risk control calculations, fund freezing, and order creation operations on order submissions, and updates the order risk control status based on the risk control judgment results. The memory recovery unit synchronously updates the backup node's memory data such as funds, positions, orders, and risk control status. Throughout the entire process, the backup node does not send any messages to the client, risk control system, or exchange order gateway.

[0064] Time-triggered operation processing: When the message sequence contains time messages and there is a time-triggered judgment requirement for data rationing, the backup node completes the rate limiting calculation based on the logical time of the time message, rather than the local time of the non-backup node. For example, if the primary node judges that the data ration within 5 seconds has not reached the threshold based on the time message t1+100 milliseconds, the backup node will also complete the same judgment based on the time message, ensuring that the calculation results of the primary and backup nodes are consistent.

[0065] Continuous synchronization: During the trading day, the master node continuously performs external message reception, time message generation and sequencing, business processing and message sending operations, while the backup node continuously receives and processes message sequences, achieving real-time and consistent synchronization of memory data between the master and backup nodes.

[0066] Example 2: Deployment of a device for primary and backup data synchronization in a securities trading system

[0067] The device in this embodiment is deployed on the server in the brokerage's data center. The master node module and the backup node module are deployed on two independent high-performance servers. The servers establish a communication link through gigabit fiber optic cables and carry a custom TCP transmission protocol. The specific deployment steps are as follows:

[0068] Master node module deployment: Deploy a message receiving unit, a time message generation unit, a sequencing unit, a business processing unit, and a message sending unit on the master server. Each unit adopts a microservice architecture for decoupled deployment and achieves data interaction through an internal message queue. Configure the external message access port of the message receiving unit to connect to the upstream client, risk control system, and exchange order gateway. Configure the generation interval of the time message generation unit to 250 milliseconds. Configure the message sending unit to asynchronous sending mode and bind it to the communication IP and port of the backup server.

[0069] Backup node module deployment: Deploy a message receiving unit, a time parsing unit, a business processing unit, and a memory recovery unit in the backup server. The software version and configuration parameters of each unit are completely consistent with those of the primary node module. Configure the communication port of the message receiving unit and establish a connection with the message sending unit of the primary node module. Close all ports of the backup node module that send messages to external systems to ensure that the backup node does not communicate with the outside world during processing.

[0070] Protocol Configuration: Configure custom TCP protocol message persistence rules and resending mechanism in the communication link between the primary and backup servers, set the message persistence storage path to the server's local solid-state drive, set the timeout for reconnection after disconnection to 500 milliseconds, and trigger resending immediately after message loss.

[0071] Joint debugging and testing: After deployment, joint debugging and testing of primary and backup nodes is carried out to simulate scenarios such as order submission, order cancellation, transaction, and risk control rule changes. This verifies the consistency between the memory data of the backup node and the primary node, and tests the message resending and data synchronization effects under abnormal scenarios such as network disconnection and message loss, to ensure that the primary and backup data synchronization is accurate and reliable in all scenarios.

[0072] Going live: After the joint debugging test is passed, the device will be connected to the actual A-share in-memory trading system of the securities firm and run on a trial basis on a trading day. After the trial run is successful, it will be officially launched to realize the synchronization of primary and backup data of the securities trading system.

[0073] Example 3: Adaptation and Application of Financial Transaction Risk Control System

[0074] The method and apparatus of this invention can be adapted to a securities firm's joint risk control system. Only the business processing units of the primary and backup node modules need to be adjusted to adapt to the business logic of the risk control system. The specific steps are as follows:

[0075] The message receiving unit of the risk control system's main node module has been adjusted to receive order messages and transaction execution report messages sent by multiple trading systems.

[0076] Adjust the business processing units of the primary and backup node modules to adapt to the joint risk control calculation logic of the risk control system, including the aggregation of orders from multiple trading systems, risk control threshold judgment, and risk warning calculation.

[0077] The generation interval of the time message generation unit remains at 100 milliseconds (to adapt to the higher time granularity requirements of the risk control system), while the configuration of other units and devices remains unchanged;

[0078] After receiving external messages, the master node of the risk control system sequences them with the time message generated on its own machine and sends them to the backup node. The backup node processes the message sequence according to the same risk control calculation logic, restores the memory data, and realizes the synchronization of master and backup data in the risk control system.

[0079] Industrial applicability

[0080] This invention discloses a method and apparatus for master-slave data synchronization in a securities trading system. It can be directly deployed in a brokerage firm's in-memory trading system. Through standardized module design and flexible configuration, it can also be quickly adapted to related financial systems such as financial trading risk control systems. This solves the core problems of existing master-slave data synchronization mechanisms being coupled with business logic and exhibiting poor consistency, significantly improving the system's high availability, stability, and development and maintenance efficiency. The technical solution of this invention does not require large-scale modifications to existing securities trading systems; it only requires adding relevant master-slave data synchronization modules to the system framework layer. This results in low modification costs, high deployment efficiency, and good industrial applicability and broad market application prospects.

[0081] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for synchronizing primary and backup data in a securities trading system, characterized in that, Includes the following steps: Step 1: Receive external input messages through the transaction node master node. The external input messages include order placement, order cancellation, transaction confirmation, and risk control system judgment results. Step 2: Generate time messages at fixed time intervals on the local machine through the master node, and sort the external input messages and time messages in a unified order to form a sorted message sequence; Step 3: The master node sends the ordered message sequence to the transaction node backup node, and the master node completes its own business logic processing based on the message sequence. Step 4: Receive the message sequence through the backup node, process the message sequence according to the same business processing logic as the master node, restore the memory data of the backup node, and the backup node does not send messages to external systems during the processing. Step 5: When the backup node is processing the message sequence, all time-triggered judgment operations are executed based on the time message in the message sequence, rather than the local time of the backup node.

2. The method according to claim 1, characterized in that, The generation interval of the time message in step 2 can be adjusted according to actual business needs. There is no fixed order between receiving external input messages and generating time messages. The sequencing process sorts the two types of messages in a unified manner.

3. The method according to claim 1, characterized in that, The method of sending message sequences from the master node to the backup node in step 3 includes asynchronous sending and synchronous sending. Asynchronous sending is used in low-latency service scenarios.

4. The method according to claim 1, characterized in that, The message transmission between the master node and the backup node in step 3 is implemented based on a protocol that supports message persistence. The protocol layer completes message resending in scenarios of disconnection or message loss.

5. The method according to claim 1, characterized in that, The process of restoring the memory data of the backup node in step 4 includes: performing risk control, fund, position and order processing within the node on the entrustment information; performing fund, position and order processing on the risk control judgment results; and performing risk control, fund, position and order processing within the node on the transaction report.

6. The method according to claim 1, characterized in that, In step 5, the backup node updates the relative time of the memory record according to the received time message, and performs memory calculations for risk control rule judgment and rate limiting judgment based on the relative time.

7. The method according to any one of claims 1-6, characterized in that, The method also includes an initialization step to ensure that the initialization state of the backup node is completely consistent with that of the master node.

8. The method according to any one of claims 1-6, characterized in that, The method is adapted to the master-slave data synchronization of financial transaction risk control systems. The master node of the risk control system receives external messages and sends them to the backup node of the risk control system after sequencing the local time message. The backup node restores the memory data according to the same logic.

9. A device for synchronizing primary and backup data in a securities trading system, used to implement the method as described in claim 1, characterized in that, The system includes a master node module and a backup node module, which are communicatively connected. The master node module includes: a message receiving unit for receiving external input messages such as order placement, order cancellation, transaction reports, and risk control system judgment results; a time message generation unit for generating time messages at fixed time intervals, which can be adjusted as needed; a sequencing unit for uniformly sequencing external input messages and time messages to form a message sequence; a business processing unit for completing the master node's own business logic processing based on the message sequence and updating the master node's memory data; and a message sending unit for... The ordered message sequence is sent to the backup node module, supporting both asynchronous and synchronous sending modes. The backup node module includes: a message receiving unit for receiving the ordered message sequence sent by the master node module; a time parsing unit for extracting time messages from the message sequence to provide logical time basis for time trigger judgment; a business processing unit for processing the message sequence according to the same business processing logic as the master node module, only restoring the backup node's memory data and not sending messages to external systems; and a memory recovery unit for updating the backup node's memory data on funds, positions, orders, and risk control status according to the business processing results.

10. The apparatus according to claim 9, characterized in that, The communication link between the master node module and the backup node module is equipped with a transmission protocol that supports message persistence. The protocol layer has the functions of reconnection after disconnection and resending lost messages. The device can be adapted to be deployed in securities trading systems and financial trading risk control systems.

Images