A high-concurrency data interface processing platform and method

Abstract

This disclosure relates to a high-concurrency data interface processing platform and method. The platform includes: an access scheduling layer for unified access to external requests, performing security protection, traffic scheduling, protocol conversion, and routing distribution; an application caching layer for supporting business applications in processing business logic, mitigating high-concurrency impacts through distributed caching and message queues, following the access order of "cached-queue-database," and asynchronously processing time-consuming operations and instantaneous peak requests from various business applications through message queues to achieve peak shaving and valley filling; and a data persistence layer for securely and efficiently storing and accessing data. This disclosure abandons the existing approach of extreme optimization of a single node, innovatively shifting to distributed system collaboration, and achieving decoupling and elasticity through layered design, thereby designing an ideal high-concurrency data interface system. Unlike existing technologies, this disclosure is not a "silver bullet" of a single technology, but a comprehensive solution balancing performance, stability, cost, and maintainability.

Description

Technical Field

[0001] This disclosure relates to the field of data interface processing, and more specifically, to a high-concurrency data interface processing platform and method. Background Technology

[0002] With the deepening of digital transformation, enterprises are increasingly demanding data interface services. In high-concurrency scenarios, data interface platforms face multi-dimensional challenges related to performance, stability, operation and maintenance, and security. Existing technologies include API management and development platforms (such as Postman, Apifox, and AWS API Gateway) focusing on the entire API lifecycle management; data integration / ETL platforms (such as FineDataLink and DataX) emphasizing the integration of heterogeneous data sources; and aggregated API marketplaces (such as RapidAPI and Aggregated Data) providing ready-made functional API services.

[0003] However, all of the above products have the following obvious limitations: (1). Low code / visualization: Platforms that can provide drag-and-drop, graphical user interfaces are relatively rare, and the technical threshold for API development and data integration is still relatively high.

[0004] (2) Chaotic API development rules: Teams cannot design and define API interface contracts before front-end and back-end development, which cannot improve collaboration efficiency.

[0005] (3) Security risks: With the improvement of data security regulations, the platform's capabilities in authentication, authorization, auditing and transmission encryption have become crucial, and most platforms cannot provide a complete security strategy.

[0006] More importantly, existing technical solutions are mostly single-function products, and no platform can simultaneously meet the comprehensive needs of API full development process management, data warehouse and ETL data interaction, and integration of common API market functions. In addition, when facing peak traffic, traditional solutions suffer from prominent problems such as high interface response latency, untimely elastic scaling, easy overload of backend services, easy failure of circuit breaker and rate limiting strategies, and difficulty in problem localization.

[0007] Currently, there is no technical solution that can manage the entire API development process, satisfy the data exchange capabilities of database warehouses and ETL, and take into account the general functions of the API market. Summary of the Invention

[0008] To address the problems existing in the prior art, this disclosure proposes a high-concurrency data interface processing platform and method to solve at least one of the aforementioned technical problems. The technical solution adopted in this disclosure is as follows: In a first aspect, this disclosure provides a high-concurrency data interface processing platform, the platform comprising: The access scheduling layer is used to unify the access of external requests and perform security protection, traffic scheduling, protocol conversion, and route distribution. The access scheduling layer includes an API gateway and a multi-level load balancing unit. The API gateway is used to integrate authentication, rate limiting, circuit breaking, logging, and traffic mirroring functions. The multi-level load balancing unit is deployed at multiple levels in the DNS layer, hardware layer, and software layer to realize intelligent distribution and fault transfer of data traffic. The application caching layer is used to support business applications in processing business logic, resist high concurrency impacts, and asynchronously process time-consuming operations and instantaneous peak requests of various business applications to achieve peak shaving and valley filling. The application caching layer includes a business application cluster, a distributed cache, and a message queue. The data persistence layer is used to securely and efficiently store and access data. The data persistence layer includes a database proxy middleware, a sharded database cluster, a master database, and multiple slave databases. The database proxy middleware is used to shield the application caching layer from the database sharding and partitioning strategies. The sharded database cluster is used to split data into multiple databases based on the sharding and partitioning strategies and business logic. The data persistence layer, based on the read-write separation strategy and the sharded database cluster, directs write operations to the master database and distributes read operations to the slave databases.

[0009] Preferably, the API gateway is the sole entry point to the platform, supports access via multiple protocols including RESTful, GraphQL, and gRPC, and is configured with dynamic routing, canary release, request anonymization, and full traffic mirroring functions.

[0010] Preferably, the API gateway works in conjunction with the multi-level load balancing unit, and when an abnormality is detected in the backend service node, it automatically triggers a circuit breaker and starts a backup node.

[0011] Preferably, the circuit breaker status is determined based on two indicators: sliding window error rate and response delay.

[0012] Preferably, the application cache layer adopts a stateless design. Under the stateless design, each business application does not participate in data persistence, thus making horizontal scaling (adding machines) easy.

[0013] Preferably, the message queue and the business application communicate between threads through a lock-free queue to avoid lock contention.

[0014] Preferably, in the distributed cache, hot data is pre-determined by a Bloom filter and then loaded into the local cache or CDN edge node.

[0015] Preferably, the stateless design is implemented by externalizing the session to a distributed cache or token service, so that any business application instance can be replaced or horizontally scaled at any time, and the scaling process does not require data migration or state synchronization.

[0016] Preferably, the database sharding strategy performs horizontal partitioning based on user ID hash, time range, or business region, and supports dynamic scaling up or down; during scaling up, a dual-write mechanism ensures data consistency, and during scaling down, a data migration service completes lossless shutdown.

[0017] Preferably, the master database and the slave database adopt a semi-synchronous replication or group replication mechanism to ensure zero data loss during failover.

[0018] Preferably, the database is configured to support multi-replica deployment, with replicas maintaining consistency via Raft or Paxos protocols.

[0019] Preferably, the database proxy middleware uses ShardingSphere, MyCat, or a self-developed middleware, supporting SQL parsing, route rewriting, result set merging, and transparent read / write separation.

[0020] Preferably, the database proxy middleware has a built-in connection pool management system that supports connection warm-up and connection leak detection.

[0021] Preferably, the platform may further include: The network acceleration layer is used to speed up network data transmission and reduce network transmission latency, and is deployed between the access scheduling layer and the external request source. The kernel bypass layer is used to bypass the operating system kernel network protocol stack through DPDK and / or RDMA technology. It is deployed between the application cache layer and the operating system to realize user-mode direct network packet processing and remote direct memory access. The monitoring and disaster recovery layer is used to integrate link tracing, log aggregation, indicator monitoring and nanosecond-level hardware timestamp measurement to achieve end-to-end observability, and runs through the access scheduling layer, application caching layer and data persistence layer.

[0022] A second aspect of this disclosure provides a method for processing high-concurrency data interfaces, the method comprising: External requests are uniformly accessed through the API gateway and intelligently distributed by a multi-level load balancing unit. The API gateway performs authentication, rate limiting, circuit breaking, and logging, and routes legitimate requests to the application cache layer. The application caching layer generates business application instances based on the received legitimate requests. The business application instances process requests in a stateless manner, prioritizing queries to the distributed cache, CDN, or client-side local cache. For data that is not cached or needs to be persisted, asynchronous peak-shaving is performed through message queues. The data persistence layer receives data access requests through a database proxy middleware. It extracts write and read operations from the data access requests and, based on a read-write separation strategy and a sharded database cluster, directs write operations to the master database and distributes read operations to slave databases.

[0023] Preferably, the sharded database cluster routes data to the sharded nodes of the corresponding database according to the database sharding and table partitioning strategy.

[0024] Preferably, semi-synchronous replication or group replication mechanisms are used to ensure data consistency between the master database and the slave database.

[0025] In a third aspect, this disclosure provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the high-concurrency data interface processing method described above.

[0026] In a fourth aspect, this disclosure provides an electronic device including a processor and a memory, the processor being configured to execute a computer program stored in the memory to implement the high-concurrency data interface processing method described above.

[0027] The beneficial effects of this disclosure are as follows: This disclosure provides a high-concurrency data interface processing platform and method, abandoning the existing approach of extreme optimization of a single node, and innovatively turning to distributed system collaboration. It achieves decoupling and elasticity through layered design, thereby designing an ideal high-concurrency data interface system. Unlike existing technologies, this disclosure is not a "silver bullet" of a single technology, but a comprehensive solution balancing performance, stability, cost, and maintainability.

[0028] This disclosure achieves a comprehensive improvement in performance, stability, and maintainability in high-concurrency scenarios through a three-layer distributed collaborative architecture. The API gateway, as the sole entry point, integrates authentication, rate limiting, circuit breaking, logging, and traffic mirroring functions. It supports multi-protocol access and dynamic routing, and, in conjunction with DNS, hardware, and software multi-level load balancing, enables intelligent distribution and failover, establishing a high-availability defense at the entry point to prevent localized failures from escalating into system avalanches. The application caching layer adopts a stateless design, enabling instance replacement and horizontal scaling at any time through externalized sessions. Following the "cache-queue-database" access order, hot data is pre-positioned to distributed cache, CDN, or client-side local cache, and loading strategies are optimized through Bloom filters. Combined with cache preheating and penetration protection mechanisms, it reduces database pressure. The message queue asynchronously smooths out time-consuming operations and instantaneous peak requests, communicating with business applications through lock-free queues to avoid lock contention, achieving traffic smoothing and system decoupling. The data persistence layer uses read-write separation to direct write operations to the master database and distribute read operations to slave databases. It performs horizontal sharding based on user ID hash or business region, supports dynamic scaling, and ensures consistency through a dual-write mechanism. The database proxy middleware transparently handles SQL parsing, route rewriting, and result set merging, so that the application caching layer does not need to worry about the underlying sharding strategy. The master and slave databases use semi-synchronous replication or group replication, and the multiple replicas of the slave databases maintain consistency through Raft or Paxos protocols, ensuring zero data loss during failover.

[0029] Furthermore, this disclosure can also add a network acceleration layer, a kernel bypass layer, and a monitoring and disaster recovery layer, forming a six-layer collaborative architecture on the original three-layer architecture. By organically integrating network layer physical optimization, system layer kernel bypass, application layer extreme efficiency, and end-to-end observability, it solves the technical problems of single function, insufficient high-concurrency processing capability, low collaboration efficiency, and incomplete security strategies in the prior art, and has significant novelty and inventiveness. Attached Figure Description

[0030] The accompanying drawings, which form part of this application, are used to provide a further understanding of this disclosure. The illustrative embodiments of this disclosure and their descriptions are used to explain this disclosure and do not constitute an undue limitation of this disclosure.

[0031] Figure 1 The architecture of the high-concurrency data interface processing platform described in Embodiment 1 of this disclosure Figure 1 .

[0032] Figure 2 This is a flowchart of the high-concurrency data interface processing method described in Embodiment 2 of this disclosure. Detailed Implementation

[0033] The present disclosure will now be described in detail with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the present application can be combined with each other.

[0034] The following detailed descriptions are exemplary and intended to provide further detailed explanation of this disclosure. Unless otherwise specified, all technical terms used in this disclosure have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this disclosure.

[0035] API Management and Development Platform: Focuses on the entire API lifecycle management, including design, development, testing, deployment, monitoring, and security. Main application scenarios: microservice governance, building open platforms, and internal collaboration within development teams. Major applications include: International / cloud vendors: Postman, Apifox, AWS API Gateway, Kong; Domestic / commercial: YAPI.

[0036] Data integration / ETL platform: Focuses on connecting and integrating multiple heterogeneous data sources, processing them, and then providing data services externally through APIs. Main application scenarios: Building data middleware and data warehouses, breaking down internal data silos within enterprises, and enabling complex data processing and API publishing. Major applications include comprehensive platforms such as FineDataLink, Informatica, and DataStage, as well as open-source / cloud-native platforms like DataX and Kafka Connect.

[0037] API Marketplace: As an "app store" for third-party APIs, it gathers a large number of ready-made, functional API services for developers to quickly call. Main application scenarios: Quickly integrate common functions such as CAPTCHAs, weather, maps, and payments, accelerating application development. Main applications include: International platforms: RapidAPI; Domestic platforms: Juhe Data, Tianxing Data, Baidu API Marketplace.

[0038] Faced with a wide range of choices in the market, users can decide which product to use based on their core needs: (1) If the user's goal is to enable the development team to collaborate more efficiently and manage the APIs they develop, the user should focus on API management and development platforms. For example, Apifox provides an integrated collaborative experience from design to testing; Postman is a popular API debugging and testing tool; and AWS API Gateway is suitable for building and managing highly available APIs on the Amazon cloud.

[0039] (2) If the user's core challenge is to extract and integrate data from various databases and SaaS systems and provide reliable data services to external parties, then a data integration / ETL platform is the user's first choice. Such platforms (such as FineDataLink or DataX) typically provide powerful connectors, visual workflow orchestration, and low-code development capabilities, which can significantly reduce the complexity of data integration.

[0040] (3) If users simply want to quickly add a general function to their application and don't want to develop from scratch, the quickest way is to find ready-made services in an aggregated API marketplace. For example, if they need SMS verification codes, they can look for aggregated data; if they need various APIs from around the world, they can check that out.

[0041] However, there is currently no product that can meet all the user's needs, namely, not only to manage the entire API development process, but also to meet the data interaction capabilities of database warehouses and ETL, and to integrate the common functions of the API marketplace.

[0042] Example 1: like Figure 1 As shown, this disclosure provides a high-concurrency data interface processing platform, the platform comprising:

[0043] The access scheduling layer 100 is used to uniformly access external requests and perform security protection, traffic scheduling, protocol conversion, and route distribution. The access scheduling layer 100 includes an API gateway 110 and a multi-level load balancing unit 120. The API gateway 110 is used to integrate authentication, rate limiting, circuit breaking, logging, and traffic mirroring functions. The multi-level load balancing unit 120 is deployed at multiple levels in the DNS layer, hardware layer, and software layer to realize intelligent distribution and fault transfer of data traffic. The application cache layer 200 is used to support business applications in processing business logic, resist high concurrency impacts, and asynchronously process time-consuming operations and instantaneous peak requests of various business applications to achieve peak shaving and valley filling; the application cache layer 200 includes a business application cluster 210, a distributed cache 220, and a message queue 230. The data persistence layer 300 is used for secure and efficient storage and access of data. The data persistence layer 300 includes a database proxy middleware 310, a sharded database cluster 320, a master database 330, and multiple slave databases 340. The database proxy middleware 310 is used to shield the application cache layer 200 from the database sharding and partitioning strategy. The sharded database cluster 320 is used to split data into multiple databases based on the database sharding and partitioning strategy and business logic. The data persistence layer 300, based on the read-write separation strategy and the sharded database cluster 320, directs write operations to the master database 330 and distributes read operations to the slave databases 340.

[0044] The collaboration relationship between the access scheduling layer 100, the application caching layer 200, and the data persistence layer 300 can be found in [reference needed]. Figure 1 To understand.

[0045] Furthermore, the API gateway 110 is the sole entry point to the platform, supports access via multiple protocols including RESTful, GraphQL, and gRPC, and is configured with dynamic routing, canary release, request desensitization, and full traffic mirroring functions.

[0046] Furthermore, the multi-level load balancing unit 120 includes: DNS global load balancer, which schedules traffic based on geographical location and data center health status; Hardware load balancers are deployed in the hosting data center and use F5 or equivalent performance devices. Software load balancers, using Nginx or LVS, support weighted round-robin, minimum number of connections, and consistent hashing strategies.

[0047] Furthermore, the API gateway 110 works in conjunction with the multi-level load balancing unit 120 to automatically trigger a circuit breaker and start a backup node when an abnormality is detected in a backend service node.

[0048] Furthermore, the circuit breaker status is determined based on two indicators: sliding window error rate and response latency.

[0049] Furthermore, the application cache layer 200 adopts a stateless design. Under the stateless design, each business application does not participate in data persistence, thus making horizontal scaling (adding machines) easy.

[0050] Furthermore, in the application cache layer 200: Business application cluster 210 integrates multiple business applications. Based on a stateless design, each business application does not participate in data persistence. Distributed caching 220 uses a Redis cluster or Memcached cluster to provide cache preheating, cache penetration protection and cache avalanche circuit breaking mechanisms for various business applications. Message queue 230 uses Kafka, RocketMQ, or Pulsar and supports ordered messages, transactional messages, and delayed messages.

[0051] Furthermore, in the application caching layer 200, the distributed cache 220 and message queue 230 are used to resist high concurrency impacts. Following the access order of "cache-queue-database", hot data is pre-positioned to the distributed cache 220, CDN or client local cache. The message queue 230 is used to asynchronously process the time-consuming operations and instantaneous peak requests of each business application, thereby achieving peak shaving and valley filling.

[0052] Furthermore, the message queue 230 and the business application communicate with each other through a lock-free queue, thus avoiding lock contention.

[0053] Furthermore, in the distributed cache 220, hot data is pre-determined by a Bloom filter and then loaded into the local cache or CDN edge node.

[0054] Furthermore, the stateless design is implemented by externalizing sessions to a distributed cache 220 or a token service, which allows any business application instance to be replaced or horizontally scaled at any time without requiring data migration or state synchronization during the scaling process.

[0055] Furthermore, the database sharding and table partitioning strategy performs horizontal partitioning based on user ID hash, time range, or business region, and supports dynamic expansion or contraction; during expansion, a dual-write mechanism ensures data consistency, and during contraction, a data migration service completes lossless shutdown.

[0056] Furthermore, the master database 330 and the slave database 340 employ a semi-synchronous replication or group replication mechanism to ensure zero data loss during failover.

[0057] Furthermore, the database 340 is configured to support multi-replica deployment, with replicas maintaining consistency via Raft or Paxos protocols.

[0058] Furthermore, the database proxy middleware 310 adopts ShardingSphere, MyCat or self-developed middleware, and supports SQL parsing, route rewriting, result set merging and transparent read-write separation; the database proxy middleware 310 has built-in connection pool management and supports connection warm-up and connection leak detection.

[0059] Furthermore, the platform may also include: The network acceleration layer is used to speed up network data transmission and reduce network transmission latency, and is deployed between the access scheduling layer 100 and the external request source. The kernel bypass layer is used to bypass the operating system kernel network protocol stack through DPDK and / or RDMA technology. It is deployed between the application cache layer 200 and the operating system to realize user-mode direct network packet processing and remote direct memory access. The monitoring and disaster recovery layer is used to integrate link tracing, log aggregation, indicator monitoring and nanosecond-level hardware timestamp measurement to achieve full-link observability, and runs through the access scheduling layer 100, application cache layer 200 and data persistence layer 300.

[0060] Furthermore, the monitoring disaster recovery layer is configured to support Active-Active dual-active architecture, deterministic replay, and millisecond-level risk control circuit breaker.

[0061] Furthermore, the network acceleration layer includes: The physical deployment optimization module deploys the core trading program and API gateway 110 in the hosting data center closest to the data source or exchange matching engine, thereby reducing transmission latency by shortening the physical distance. The ultra-low latency network protocol module uses InfiniBand or RoCE protocols to replace the traditional TCP / IP protocol in the internal data center, enabling remote direct memory access and kernel bypass transmission. The hardware acceleration module, using FPGA or smart network interface card, offloads network protocol parsing, data packet verification, risk control rule matching, and transaction logic from the CPU to hardware execution, achieving nanosecond-level processing speed.

[0062] Furthermore, the kernel bypass layer includes: The DPDK user-space network processing module, driven by polling mode and large page memory mapping, directly processes network packets in user space, completely bypassing the operating system kernel network protocol stack and eliminating interrupt context switching overhead. The RDMA (Remote Direct Memory Access) module allows one host to directly access the memory space of another host without the need for the other host's CPU to participate in data copying, achieving zero-copy ultra-low latency communication. The CPU and memory optimization module includes: a core binding and isolation unit, which binds critical processes to specific CPU cores and disables interrupt response for those cores, ensuring that critical processes have exclusive access to computing resources; a large page memory management unit, which uses HugePages to reduce the number of TLB misses; and a NUMA awareness unit, which enables the CPU to prioritize accessing local memory nodes and avoid cross-node access latency.

[0063] Furthermore, the application cache layer 200 also includes an application layer extreme optimization module, which includes: The simplified protocol encoding and decoding unit uses a binary protocol instead of the JSON / XML text protocol. Fields are arranged with fixed offsets and compact layout. The encoding and decoding process does not require lexical analysis and is completed directly through memory copying and bit operations. High-efficiency data structure units, using arrays or vectors for contiguous memory storage throughout, avoiding CPU cache invalidation caused by linked list pointer jumps; inter-thread communication uses lock-free circular queues, and producer-consumer synchronization is achieved based on atomic operations; The memory pre-allocation unit pre-allocates all runtime resources through a memory pool when the system starts up. Dynamic memory allocation and system calls are prohibited on the core processing path to avoid the uncertain delays and memory fragmentation caused by malloc / free.

[0064] Furthermore, the monitoring disaster recovery layer includes: The nanosecond-level monitoring unit uses hardware timestamps and the PTP precise time protocol to measure the end-to-end full-link latency from receiving a request to responding and leaving the site at the nanosecond level, and records the latency contribution value of each link. Active-Active dual-active disaster recovery unit deploys peer active nodes in multiple geographical regions. Each node provides services to the outside world simultaneously and maintains eventual consistency through a data synchronization layer. When any node fails, traffic is switched to the live node within seconds. The deterministic replay unit records all external inputs and internal state change events to a persistent log. During fault recovery, the system state can be quickly reconstructed and the root cause of the fault can be located by accurately replaying the log. The risk control circuit breaker unit sets hard delay thresholds and business loss thresholds. When the monitoring indicators trigger the thresholds, the system automatically cuts off the order flow or request flow within milliseconds to prevent the spread of faults and catastrophic losses.

[0065] Example 2: like Figure 2 As shown, in a second aspect, this disclosure provides a high-concurrency data interface processing method, the method comprising: S100. External requests are uniformly accessed through API gateway 110 and intelligently distributed by multi-level load balancing unit 120. API gateway 110 performs authentication, rate limiting, circuit breaking and logging, and routes legitimate requests to application cache layer 200. S200 and the application cache layer 200 generate business application instances based on the received legitimate requests. The business application instances process requests in a stateless manner, prioritizing queries to the distributed cache 220, CDN, or client-side local cache. For data that is not cached or needs to be persisted, asynchronous peak-shaving is performed through the message queue 230. S300 and the data persistence layer 300 receive data access requests through the database proxy middleware 310. The data persistence layer 300 extracts write and read operations from the data access requests. Based on the read-write separation strategy and the sharded database cluster 320, the write operations are directed to the master database 330 and the read operations are distributed to the slave database 340.

[0066] Furthermore, the sharded database cluster 320 routes data to the corresponding sharded nodes of the database according to the database sharding and table partitioning strategy.

[0067] Furthermore, semi-synchronous replication or group replication mechanisms are used to ensure data consistency between the master database 330 and the slave database 340.

[0068] Furthermore, for requests with extremely high real-time requirements, network data packets are processed directly by bypassing the operating system kernel protocol stack through a kernel bypass layer.

[0069] Furthermore, during the execution of the method, full-link observability is achieved by monitoring the disaster recovery layer's distributed link tracing, log aggregation, and indicator monitoring; end-to-end latency is measured using nanosecond-level hardware timestamps; when a node failure or performance threshold exceeding the limit is detected, Active-Active dual-active switching or millisecond-level risk control circuit breaker is triggered; and resource scale is automatically adjusted to match real-time traffic through elastic scaling units.

[0070] Example 3: Embodiment 3 of this disclosure provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the high-concurrency data interface processing method as described in Embodiment 2.

[0071] The computer-readable storage medium includes volatile or non-volatile, removable or non-removable media implemented in any platform or technology for storing information (such as computer-readable instructions, data structures, computer program modules, or other data). Computer-readable storage media include, but are not limited to, RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory or other memory technologies, CD-ROM (Compact Disc Read-Only Memory), DVD or other optical disc storage, cartridges, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer.

[0072] Example 4: Embodiment 4 of this disclosure provides an electronic device including a processor and a memory, wherein the processor is used to execute a computer program stored in the memory to implement the high-concurrency data interface processing method described in Embodiment 1.

[0073] Based on this understanding, all or part of the processes in the platform of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, it can implement the steps of the various platform embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, and read-only memory (ROM).

[0074] Those skilled in the art will understand that embodiments of this disclosure can be provided as a platform, method, or computer program product. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0075] This disclosure is described with reference to flowchart illustrations and / or block diagrams of platforms, apparatus (methods), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a mechanism for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0076] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0077] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0078] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0079] In summary, the high-concurrency data interface processing platform and method provided in embodiments 1-4 of this disclosure achieve a comprehensive improvement in performance, stability, and maintainability in high-concurrency scenarios through a three-layer distributed collaborative architecture. The API gateway, as the sole entry point, integrates authentication, rate limiting, circuit breaking, logging, and traffic mirroring functions, supports multi-protocol access and dynamic routing, and, in conjunction with DNS, hardware, and software multi-level load balancing, achieves intelligent distribution and failover, establishing a high-availability defense at the entry point to prevent localized failures from spreading into system avalanches. The application caching layer adopts a stateless design, enabling instance replacement and horizontal scaling at any time through externalized sessions; following the "cachrystal-queue-database" access order, hot data is pre-positioned to the distributed cache 220, CDN, or client-side local cache, and the loading strategy is optimized through a Bloom filter, combined with cache preheating and penetration protection mechanisms to reduce database pressure; the message queue 230 asynchronously smooths out time-consuming operations and instantaneous peak requests, communicating with business applications through lock-free queues to avoid lock contention, achieving traffic smoothing and system decoupling. The data persistence layer uses read-write separation to direct write operations to the master database and distribute read operations to slave databases. It performs horizontal sharding based on user ID hash or business region, supports dynamic scaling, and ensures consistency through a dual-write mechanism. The database proxy middleware transparently handles SQL parsing, route rewriting, and result set merging, so that the application caching layer does not need to worry about the underlying sharding strategy. The master and slave databases use semi-synchronous replication or group replication, and the multiple replicas of the slave databases maintain consistency through Raft or Paxos protocols, ensuring zero data loss during failover.

[0080] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure and not to limit them. Although this disclosure has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of this disclosure. Any modifications or equivalent substitutions that do not depart from the spirit and scope of this disclosure should be covered within the protection scope of the claims of this disclosure.

Claims

1. A high-concurrency data interface processing platform, characterized in that, The platform includes: The access scheduling layer (100) is used to unify access to external requests and perform security protection, traffic scheduling, protocol conversion and route distribution; the access scheduling layer (100) includes an API gateway (110) and a multi-level load balancing unit (120); the API gateway (110) is used to integrate authentication, rate limiting, circuit breaking, logging and traffic mirroring functions; the multi-level load balancing unit (120) is deployed at multiple levels in the DNS layer, hardware layer and software layer to realize intelligent distribution and fault transfer of data traffic; The application cache layer (200) is used to support business applications in processing business logic, resist high concurrency impacts, and asynchronously process time-consuming operations and instantaneous peak requests of various business applications to achieve peak shaving and valley filling; the application cache layer (200) includes a business application cluster (210), a distributed cache (220) and a message queue (230). The data persistence layer (300) is used to securely and efficiently store and access data. The data persistence layer (300) includes a database proxy middleware (310), a sharded database cluster (320), a master database (330), and multiple slave databases (340). The database proxy middleware (310) is used to shield the application cache layer (200) from the database sharding and partitioning strategy. The sharded database cluster (320) is used to split data into multiple databases based on the sharding and partitioning strategy and business logic. The data persistence layer (300) directs write operations to the master database (330) and distributes read operations to the slave databases (340) based on the read-write separation strategy and the sharded database cluster (320).

2. The high-concurrency data interface processing platform as described in claim 1, characterized in that, The API gateway (110) is the only entry point of the platform, supports access via multiple protocols including RESTful, GraphQL and gRPC, and is configured with dynamic routing, canary release, request desensitization and full traffic mirroring functions.

3. The high-concurrency data interface processing platform as described in claim 1, characterized in that, The multi-level load balancing unit (120) includes: DNS global load balancer, which schedules traffic based on geographical location and data center health status; Hardware load balancers are deployed in the hosting data center and use F5 or equivalent performance devices. Software load balancers, using Nginx or LVS, support weighted round-robin, minimum number of connections, and consistent hashing strategies.

4. The high-concurrency data interface processing platform as described in claim 1, characterized in that, The API gateway (110) works in conjunction with the multi-level load balancing unit (120) to automatically trigger a circuit breaker and start a backup node when an abnormality is detected in the backend service node. The circuit breaker status is determined based on two indicators: sliding window error rate and response latency.

5. The high-concurrency data interface processing platform as described in claim 1, characterized in that, The application cache layer (200) adopts a stateless design, under which each business application does not participate in data persistence.

6. The high-concurrency data interface processing platform as described in claim 1, characterized in that, In the application cache layer (200): The business application cluster (210) integrates multiple business applications. Based on a stateless design, each business application does not participate in data persistence. Distributed caching (220) uses Redis cluster or Memcached cluster to provide cache preheating, cache penetration protection and cache avalanche circuit breaking mechanism for various business applications; The message queue (230) uses Kafka, RocketMQ or Pulsar and supports sequential messages, transactional messages and delayed messages; the message queue (230) and the business application communicate with each other through a lock-free queue to avoid lock contention. Among them, the distributed cache (220) and message queue (230) are used to resist the impact of high concurrency. The access order of "cache-queue-database" is followed, and hot data is placed in front of the distributed cache (220), CDN or client local cache. The time-consuming operations and instantaneous peak requests of each business application are made asynchronous through the message queue (230) to achieve peak shaving and valley filling.

7. The high-concurrency data interface processing platform as described in claim 1, characterized in that, The database sharding and table partitioning strategy performs horizontal partitioning based on user ID hash, time range, or business region, and supports dynamic scaling up or down. During scaling up, a dual-write mechanism ensures data consistency, and during scaling down, a data migration service completes lossless shutdown. The master database (330) and the slave database (340) adopt a semi-synchronous replication or group replication mechanism to ensure zero data loss during failover.

8. The high-concurrency data interface processing platform as described in claim 1, characterized in that, The database proxy middleware (310) adopts ShardingSphere, MyCat or self-developed middleware, and supports SQL parsing, route rewriting, result set merging and transparent read-write separation.

9. The high-concurrency data interface processing platform as described in claim 1, characterized in that, The platform also includes: A network acceleration layer, used to speed up network data transmission and reduce network transmission latency, is deployed between the access scheduling layer (100) and the external request source; The kernel bypass layer is used to bypass the operating system kernel network protocol stack through DPDK and / or RDMA technology. It is deployed between the application cache layer (200) and the operating system to realize user-mode direct network packet processing and remote direct memory access. The monitoring and disaster recovery layer is used to integrate link tracing, log aggregation, indicator monitoring and nanosecond-level hardware timestamp measurement to achieve full-link observability, and runs through the access scheduling layer (100), application caching layer (200) and data persistence layer (300).

10. A high-concurrency data interface processing method, characterized in that, The method includes: External requests are uniformly accessed through the API gateway (110) and intelligently distributed by the multi-level load balancing unit (120); the API gateway (110) performs authentication, rate limiting, circuit breaking and logging, and routes legitimate requests to the application cache layer (200); The application caching layer (200) generates business application instances based on the received legitimate requests. The business application instances process requests in a stateless manner, prioritizing queries to the distributed cache (220), CDN, or client local cache. For data that is not cached or needs to be persisted, asynchronous peak shaving is performed through a message queue (230). The data persistence layer (300) receives data access requests through the database proxy middleware (310). The data persistence layer (300) extracts write and read operations from the data access requests. Based on the read-write separation strategy and the sharded database cluster (320), the write operations are directed to the master database (330) and the read operations are distributed to the slave databases (340). The sharded database cluster (320) routes the data to the sharded nodes of the corresponding database according to the database sharding and table partitioning strategy.

Images