CDP-based data fusion method and system

Abstract

The invention relates to a CDP-based data fusion method and system. Comprising a multi-source heterogeneous data real-time acquisition and preprocessing module, a cross-platform identity recognition module based on a graph neural network, a dynamic label management and customer portrait construction module, a privacy protection and data security control module, a continuous data protection and rapid recovery module and a data service and application interface module, customer data is collected from a multi-source heterogeneous system through a self-adaptive data connector, real-time and batch processing is performed by adopting a stream batch integrated processing engine, cross-platform customer identity recognition is realized based on a graph neural network, customer tags are generated and managed through a dynamic tag management system, and data security is protected by applying a privacy computing technology. And realizing data tracking and recovery by utilizing a continuous data protection mechanism. According to the method, the federated learning framework, the real-time stream processing engine, the dynamic label management system and the privacy computing technology are introduced, so that the data fusion efficiency, the real-time performance and the compliance of the CDP system are remarkably improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of data fusion methods, in particular to a data fusion method based on CDP and a system thereof. BACKGROUND

[0002] A plurality of PDPs are merged into one CDP in the RRDtool tool. That is, the result after performing the merge statistics function operation. This value is the data stored in the RRA, and these data are also used when drawing. The data fusion method of the CDP mainly includes data acquisition, integration, cleaning and deduplication processing steps. The CDP platform supports the access of multiple data sources, including websites, APPs, mini programs, databases, social media, emails and multiple channel sources. Through these steps, the CDP can quickly and accurately integrate various data and build a perfect user portrait to provide comprehensive customer insight information for enterprises.

[0003] The customer data platform (CDP) as the core infrastructure of enterprise digital transformation has received widespread attention in recent years. The traditional CDP system mainly faces the following technical bottlenecks:

[0004] 1. Data island problem: Enterprise customer data is scattered in CRM, ERP, e-commerce platform, social media and other independent systems, and lacks an effective integration mechanism.

[0005] 2. Lack of real-time performance: Most CDP systems use batch processing, and data updates are delayed seriously, which cannot support real-time marketing decisions.

[0006] 3. Privacy compliance risks: With the implementation of the Personal Information Protection Law and other regulations, traditional CDPs have compliance risks in data collection, storage and use.

[0007] 4. Difficulty in cross-platform identity recognition: Although the ID-Mapping technology can solve some problems, it is not accurate in complex scenarios such as device number encryption and anonymous users.

[0008] In view of the above problems, the application provides an innovative data fusion method based on CDP to solve the above problems. SUMMARY

[0009] In view of the deficiencies of the prior art, the application provides a data fusion method based on CDP, which introduces a federated learning framework, a real-time stream processing engine, a dynamic label management system and a privacy computing technology, significantly improving the data fusion efficiency, real-time performance and compliance of the CDP system, and solving the problems of data island, lack of real-time performance, privacy compliance risks and difficulty in cross-platform identity recognition faced by traditional CDP systems.

[0010] To achieve the above object, the present application provides the following technical scheme: A data fusion method based on CDP, comprising the steps of:

[0011] S1, collecting customer data from multiple source heterogeneous systems through an adaptive data connector;

[0012] S2, using a stream batch integrated processing engine for real-time and batch processing;

[0013] S3, realizing cross-platform customer identity recognition based on a graph neural network;

[0014] S4, generating and managing customer labels through a dynamic label management system;

[0015] S5, applying privacy computing technology to protect data security;

[0016] S6, realizing data tracking and recovery by using a continuous data protection mechanism.

[0017] Further, the identity recognition step comprises:

[0018] 1) extracting user identifiers from each data source to construct an initial identity graph;

[0019] 2) learning node embedding representation through a graph convolution network;

[0020] 3) finding potential associated identifiers based on similarity;

[0021] 4) verifying the validity of the association in combination with business rules.

[0022] Further, the label management step comprises:

[0023] 1) defining basic labels, derived labels and predicted labels;

[0024] 2) monitoring label usage frequency, accuracy and business value indicators;

[0025] 3) making optimization or elimination suggestions for inefficient labels;

[0026] 4) retaining label history versions to support backtracking and auditing.

[0027] Further, the privacy protection step comprises:

[0028] 1) applying TLS1.3 encryption to transmission data;

[0029] 2) using AES-256 encryption to store static data;

[0030] 3) implementing homomorphic encryption on sensitive fields;

[0031] 4) adding differential privacy noise in statistical analysis.

[0032] Further, the data protection step includes:

[0033] 1) Capture data changes through database log parsing;

[0034] 2) Record business operation context information;

[0035] 3) Generate incremental snapshot optimized storage space;

[0036] 4) Provide two recovery modes of time point and business scenario.

[0037] A system of a CDP-based data fusion method, comprising a multi-source heterogeneous data real-time acquisition and preprocessing module, a cross-platform identity recognition module based on a graph neural network, a dynamic label management and customer portrait construction module, a privacy protection and data security control module, a continuous data protection and rapid recovery module, and a data service and application interface module;

[0038] The multi-source heterogeneous data real-time acquisition and preprocessing module comprises:

[0039] An adaptive data connector supports multiple data access methods such as API, ETL, burying points, and logs;

[0040] A stream-batch integrated processing engine realizes the unification of real-time stream computing and batch processing;

[0041] An intelligent data quality detection unit automatically identifies and repairs dirty data;

[0042] The cross-platform identity recognition module based on the graph neural network constructs a multi-layer identity graph containing device ID, cookie, mobile phone number, and WeChat ID, adopts a graph convolution network (GCN) to learn the association mode between identifiers, and introduces an attention mechanism to realize dynamic weight adjustment;

[0043] The dynamic label management and customer portrait construction module adopts a three-layer architecture:

[0044] A basic label layer directly extracts factual labels from raw data;

[0045] A derived label layer calculates rule-based labels based on business rules;

[0046] A predicted label layer generates predicted labels through a machine learning model;

[0047] The privacy protection and data security control module comprises:

[0048] A full-stack encryption scheme covers the transmission layer, storage layer, and computing layer;

[0049] A homomorphic encryption unit supports statistical operations in an encrypted state;

[0050] Differential privacy component, adding controllable noise in statistical analysis;

[0051] The continuous data protection and rapid recovery module captures data changes by parsing database logs and applying events, uses incremental snapshot technology, only stores changed data blocks, supports columnar compression and block-level deduplication to optimize storage efficiency;

[0052] The data service and application interface module provides real-time crowd circle selection API, supports multi-condition combined query, customer portrait query service, returns 360-degree customer view, marketing effect analysis tool, evaluates activity ROI, data export and report generation functions.

[0053] Further, the adaptive data connector can automatically identify structured, semi-structured and unstructured data sources, dynamically adjust data collection frequency and batch size, and provide a visual field mapping configuration interface.

[0054] Further, the cross-platform identity recognition module based on graph neural network further comprises:

[0055] Online inference engine, real-time calculation of embedding vector of new identifier;

[0056] Correlation verification unit, combined with business rules to verify recognition results;

[0057] Model optimization component, continuously monitors and improves recognition accuracy;

[0058] The dynamic label management and customer portrait construction module further comprises:

[0059] Label life cycle management unit, monitoring label usage frequency and business value;

[0060] Label version control system, records historical changes of definition and calculation logic;

[0061] Visual label configuration interface, supporting drag-and-drop rule definition.

[0062] Further, the privacy protection module further comprises:

[0063] Hybrid access control system based on RBAC and ABAC;

[0064] Dynamic access strategy engine, considering multi-dimensional factors such as time and location;

[0065] Automated compliance audit tool, generating privacy impact assessment report;

[0066] The continuous data protection and rapid recovery module further comprises:

[0067] Time point recovery function, accurate to seconds of data rollback;

[0068] Business scenario recovery function, semantic-based integrity recovery;

[0069] Sandbox test environment, verify the recovery effect before going online.

[0070] Compared with the prior art, the technical scheme of the application has the following beneficial effects:

[0071] 1. The application improves data fusion efficiency: through the adaptive data connector and the stream batch integrated processing engine, the data access and processing efficiency is improved by 3-5 times, meeting the real-time marketing scene demand.

[0072] 2. The application enhances customer identification accuracy: the accuracy of the GNN-based identity recognition system in the cross-platform scene is improved by 20-30% compared with the traditional method, significantly reducing marketing resource waste.

[0073] 3. The application optimizes label management: the dynamic label system improves the label usage rate by more than 40%, reducing the impact of "zombie labels" on system performance.

[0074] 4. The application ensures privacy compliance: a variety of privacy protection technologies are used comprehensively to help enterprises meet international privacy regulations such as GDPR, CCPA, etc., reducing compliance risks.

[0075] 5. The application strengthens data resilience: the improved CDP mechanism shortens the system recovery time target (RTO) to minutes, the recovery point target (RPO) is close to zero, and the business continuity assurance capability is greatly improved.

[0076] 6. The application reduces operating costs: intelligent data quality detection and label optimization functions can reduce 30-50% of manual maintenance workload. BRIEF DESCRIPTION OF DRAWINGS

[0077] Figure 1 Flowchart of the data fusion method based on CDP of the application;

[0078] Figure 2 Block diagram of the data fusion system based on CDP of the application. DETAILED DESCRIPTION

[0079] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0080] Please refer to Figure 1The CDP-based data fusion method in the embodiment includes the following steps:

[0081] S1, collecting customer data from multiple source heterogeneous systems through an adaptive data connector;

[0082] S2, using a stream-batch integrated processing engine for real-time and batch processing;

[0083] S3, realizing cross-platform customer identity recognition based on a graph neural network;

[0084] S4, generating and managing customer labels through a dynamic label management system;

[0085] S5, applying privacy computing technology to protect data security;

[0086] S6, realizing data tracking and recovery by using a continuous data protection mechanism.

[0087] In the embodiment, user identifiers are extracted from each data source to construct an initial identity graph during the identity recognition process, node embedding representation is learned through a graph convolution network, potential associated identifiers are found based on similarity, and the associated validity is verified in combination with business rules.

[0088] Specifically, the label management first defines basic labels, derived labels and predicted labels, then monitors the label usage frequency, accuracy and business value indicators, then proposes optimization or elimination suggestions for inefficient labels, and finally retains label history versions to support backtracking and auditing.

[0089] In the embodiment, in the privacy protection, TLS1.3 encryption is applied to transmitted data, AES-256 encryption is used for static data storage, homomorphic encryption is implemented for sensitive fields, and differential privacy noise is added in statistical analysis.

[0090] In the embodiment, the data protection captures data changes through database log analysis, records business operation context information, generates incremental snapshots to optimize storage space, and provides two recovery modes of time point and business scenario.

[0091] Please refer to Figure 2 The system of the CDP-based data fusion method in the embodiment includes a multi-source heterogeneous data real-time collection and preprocessing module, a cross-platform identity recognition module based on a graph neural network, a dynamic label management and customer portrait construction module, a privacy protection and data security control module, a continuous data protection and rapid recovery module, and a data service and application interface module.

[0092] The multi-source heterogeneous data real-time collection and preprocessing module includes:

[0093] An adaptive data connector supports multiple data access methods such as API, ETL, burying points and logs.

[0094] A stream-batch integrated processing engine realizes the unification of real-time stream computing and batch processing.

[0095] An intelligent data quality detection unit automatically identifies and repairs dirty data.

[0096] The cross-platform identity recognition module based on the graph neural network constructs a multi-layer identity graph containing device ID, cookie, mobile phone number, and WeChat ID, learns the association pattern between identifiers using a graph convolution network (GCN), and introduces an attention mechanism to realize dynamic weight adjustment.

[0097] The dynamic label management and customer portrait construction module adopts a three-layer architecture:

[0098] The basic label layer directly extracts factual labels from raw data, can automatically identify structured, semi-structured, and unstructured data sources, dynamically adjusts data collection frequency and batch size, and provides a visual field mapping configuration interface.

[0099] The derived label layer calculates rule-based rule labels.

[0100] The prediction label layer generates prediction labels through machine learning models.

[0101] The privacy protection and data security control module includes:

[0102] A full-stack encryption scheme covers the transmission layer, storage layer, and computing layer.

[0103] A homomorphic encryption unit supports statistical operations in an encrypted state.

[0104] A differential privacy component adds controllable noise in statistical analysis.

[0105] The continuous data protection and rapid recovery module captures data changes by analyzing database logs and application events, uses incremental snapshot technology to store only changed data blocks, and supports columnar compression and block-level deduplication to optimize storage efficiency.

[0106] The data service and application interface module provides real-time people circle selection API, supports multi-condition combined query, customer portrait query service, returns 360-degree customer view, marketing effect analysis tool, evaluates activity ROI, and data export and report generation functions.

[0107] In this embodiment, the cross-platform identity recognition module based on the graph neural network further includes:

[0108] An online inference engine calculates the embedding vector of a new identifier in real time.

[0109] An association verification unit verifies the recognition result in combination with business rules.

[0110] Model optimization component, continuously monitors and improves identification accuracy;

[0111] In this embodiment, the dynamic label management and customer portrait construction module further comprises:

[0112] Label lifecycle management unit, monitoring label usage frequency and business value;

[0113] Label version control system, record the history change of definition and calculation logic;

[0114] Visual label configuration interface, support drag-and-drop rule definition.

[0115] In this embodiment, the privacy protection module further comprises a hybrid access control system based on RBAC and ABAC, a dynamic access strategy engine considering multi-dimensional factors such as time and location, an automated compliance audit tool, and a privacy impact assessment report generator;

[0116] The continuous data protection and rapid recovery module further comprises a time point recovery function, a data rollback function accurate to seconds, a business scenario recovery function, a semantic-based integrity recovery function, a sandbox test environment, and a recovery effect verification function before going online.

[0117] Embodiment 1: Implementation of CDP system in retail e-commerce industry

[0118] 1. System deployment environment

[0119] 1.1. Hardware configuration:

[0120] Computing nodes: 8 Dell PowerEdge R750 servers, each configured with dual Intel Xeon Gold 6330 processors (28 cores / 56 threads) and 512GB of memory.

[0121] Storage system: Dell EMC PowerStore 1000T all-flash array, configured with 50TB of available capacity.

[0122] Network equipment: Cisco Nexus 93180YC-EX switch, 40GbE network connection.

[0123] 1.2. Software environment:

[0124] Operating system: Red Hat Enterprise Linux 8.4.

[0125] Container platform: Kubernetes 1.23 cluster.

[0126] Database: MongoDB 5.0 sharded cluster + Apache Cassandra 4.0.

[0127] Stream processing: Apache Flink 1.14 + Apache Kafka 3.1.

[0128] 2. Data collection and implementation

[0129] 2.1. E-commerce platform data access: We configure a JDBC source connector through Kafka Connect to capture binlog changes to the MySQL order table in real time. Tracked data is collected using a self-developed SDK, processing a peak of 120,000 events per second. Log files are collected using Filebeat and imported into the Elasticsearch cluster.

[0130] 2.2. Third-party data integration: WeChat Mini Program data is synchronized once an hour through the open platform API, and advertising platform data is obtained as CSV files via SFTP on a daily basis.

[0131] Implementation case: A clothing brand connected to 15 data sources and processed an average of 1.2 TB of data per day.

[0132] 3. Identity Recognition Implementation

[0133] 3.1. Graph Construction: Initially load 20 million user nodes and 180 million associated edges, use the DGL framework to build a graph neural network, and configure a 3-layer GCN.

[0134] Training parameters: learning rate 0.001, batch size 1024, epochs 50.

[0135] 3.2. Real-time matching: When a new user visits, similarity calculation is completed within 50ms

[0136] Implementation effect: The cross-device recognition accuracy rate reached 92.3%, an increase of 27% over traditional methods.

[0137] 4. Tag management implementation

[0138] 4.1. Tag Factory Configuration: 86 basic tags are pre-set (such as gender, region, and purchase frequency), and business personnel can independently create derivative tags such as "promotion-sensitive users":

[0139] CASE

[0140] WHEN Historical participation in promotions > 3 AND Promotion consumption share > 40% THEN 'Gao Min'

[0141] WHEN Historical participation in promotions > 1 THEN 'Medium Sensitive'

[0142] ELSE' low sensitivity'

[0143] END

[0144] Machine learning model generates prediction label "Double 11 purchase probability", AUC reaches 0.83.

[0145] 4.2, Life cycle management: the system automatically marks the label that has not been used for 3 months, generates a label health report every month, and suggests optimization solutions.

[0146] 5, Privacy protection implementation

[0147] 5.1, Encryption scheme: user mobile phone uses FPE format to keep encryption, and payment amount field uses Paillier homomorphic encryption.

[0148] Implementation cost: the performance loss of encryption and decryption is controlled within 15%.

[0149] 5.2, Access control: define 12 roles and 156 fine-grained permission rules, and dynamically verify OTP code for sensitive data access.

[0150] 6, Data recovery implementation

[0151] 6.1, Change capture: use Debezium to capture MySQL changes, process 8000+ events per second, and add business operation context through Kafka messages.

[0152] 6.2, Recovery drill: simulate order table deletion, complete recovery in 3 minutes and 12 seconds, and find and fix 2 data consistency problems during sandbox verification.

[0153] Example 2: Implementation of bank customer 360 view

[0154] 1, Special configuration

[0155] 1.1, Compliance enhancement: deploy hardware security module (HSM) to manage encryption keys, and synchronize audit logs to blockchain storage in real time.

[0156] 1.2 Risk label: develop anti-fraud combined label, integrate transaction behavior anomaly degree (LSTM model), device fingerprint similarity, and geographic location rationality.

[0157] Implementation effect: fraud identification rate increased by 33%, false positive rate decreased by 18%.

[0158] 2, System integration

[0159] 2.1, Core banking system interface: get account change information through IBM MQ in real time and load customer asset data in batches every day.

[0160] 2.2, Marketing system linkage: Real-time API response time P99 < 200ms, supporting 1500 portrait queries per second.

[0161] Embodiment 3: Customer journey management in the automotive industry

[0162] 1. Special processing

[0163] 1.1, Long-period data management: Configure special retention strategy: test drive data retention for 5 years, realize version tracking of changes in car purchase intention.

[0164] 1.2, Cross-channel integration: Exhibition visitor identification: integrate WiFi probe + CRM reservation data, and conduct correlation analysis of online behavior and offline services.

[0165] 2. Effect evaluation

[0166] 2.1. Business indicators: Sales lead conversion rate increased by 22%, customer retention cost decreased by 35%, and personalized recommendation acceptance rate reached 68%.

[0167] 2.2 System performance: Under 20 million customer data, label calculation time is 4.7 hours, and identity recognition service TP99 delay < 300ms.

[0168] In summary, the present application improves the data access and processing efficiency by 3-5 times through adaptive data connectors and stream-batch integrated processing engines, meeting the needs of real-time marketing scenarios. The identity recognition system based on GNN has an accuracy rate that is 20-30% higher than traditional methods in cross-platform scenarios, significantly reducing marketing resource waste. The dynamic label system improves label usage rate by more than 40%, reducing the impact of "zombie labels" on system performance. The comprehensive use of various privacy protection technologies helps enterprises meet international privacy regulations such as GDPR and CCPA, reducing compliance risks. The improved CDP mechanism shortens the system recovery time objective (RTO) to minutes and the recovery point objective (RPO) to near zero, significantly improving business continuity assurance capabilities. Intelligent data quality detection and label optimization functions can reduce 30-50% of manual maintenance workload.

[0169] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0170] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A data fusion method based on CDP, characterized in that: Including steps: S1. Collect customer data from multiple heterogeneous systems through adaptive data connectors; S2, using a stream-batch integrated processing engine for real-time and batch processing; S3, cross-platform customer identity recognition based on graph neural network; S4. Generate and manage customer tags through a dynamic tag management system; S5. Apply privacy computing technology to protect data security; S6. Use continuous data protection mechanism to achieve data tracking and recovery.

2. The data fusion method based on CDP according to claim 1, characterized in that: The identity identification step includes: 1) Extract user identifiers from each data source to build an initial identity graph; 2) Learning node embedding representations through graph convolutional networks; 3) Find potential associated identifiers based on similarity; 4) Verify the validity of the association based on business rules.

3. The data fusion method based on CDP according to claim 1, characterized in that: The tag management step includes: 1) Define basic labels, derived labels, and predicted labels; 2) Monitor tag usage frequency, accuracy, and business value indicators; 3) Propose optimization or elimination suggestions for inefficient tags; 4) Preserve historical versions of tags to support backtracking and auditing.

4. The data fusion method based on CDP according to claim 1, characterized in that: The privacy protection steps include: 1) Apply TLS 1.3 encryption to transmitted data; 2) Use AES-256 encryption to store static data; 3) Implement homomorphic encryption for sensitive fields; 4) Add differential privacy noise in statistical analysis.

5. The data fusion method based on CDP according to claim 1, characterized in that: The data protection steps include: 1) Capture data changes through database log analysis; 2) Record business operation context information; 3) Generate incremental snapshots to optimize storage space; 4) Provides two recovery modes: point-in-time recovery and business scenario recovery.

6. A system for the CDP-based data fusion method according to any one of claims 1 to 5, characterized in that: It includes a multi-source heterogeneous data real-time collection and pre-processing module, a cross-platform identity recognition module based on graph neural networks, a dynamic tag management and customer profile building module, a privacy protection and data security control module, a continuous data protection and rapid recovery module, and a data service and application interface module; The multi-source heterogeneous data real-time acquisition and preprocessing module includes: Adaptive data connector, supporting multiple data access methods including API, ETL, tracking data, and logs; Stream-batch integrated processing engine, unifying real-time stream computing and batch processing; Intelligent data quality detection unit, automatically identifying and repairing dirty data; The cross-platform identity recognition module based on graph neural networks builds a multi-layer identity graph containing device ID, cookies, mobile phone number, and WeChat ID. It uses a graph convolutional network (GCN) to learn the association pattern between identifiers and introduces an attention mechanism to achieve dynamic weight adjustment. The dynamic tag management and customer profile building module adopts a three-tier architecture: The basic label layer extracts factual labels directly from the raw data; Derived tag layer, which calculates rule-based tags based on business rules; Prediction label layer, which generates predictive labels through machine learning models; The privacy protection and data security control module includes: Full-stack encryption solution, covering the transport layer, storage layer, and computing layer; Homomorphic encryption unit, supporting statistical operations in encrypted state; A differential privacy component that adds controllable noise to statistical analyses; The continuous data protection and rapid recovery module captures data changes by parsing database logs and application events, uses incremental snapshot technology to store only changed data blocks, and supports columnar compression and block-level deduplication to optimize storage efficiency; The data service and application interface module provides a real-time crowd selection API, supports multi-condition combination query, customer portrait query service, returns a 360-degree customer view, marketing effect analysis tools, evaluates activity ROI, and has data export and report generation functions.

7. The CDP-based data fusion system according to claim 6, characterized in that: The adaptive data connector can automatically identify structured, semi-structured and unstructured data sources, dynamically adjust data collection frequency and batch size, and provide a visual field mapping configuration interface.

8. The CDP-based data fusion system according to claim 6, characterized in that: The cross-platform identity recognition module based on graph neural network also includes: Online inference engine, which calculates embedding vectors of new identifiers in real time; The associated verification unit verifies the recognition results in combination with business rules; Model optimization components to continuously monitor and improve recognition accuracy; The dynamic tag management and customer profile building module also includes: Tag lifecycle management unit, which monitors tag usage frequency and business value; Tag version control system to record historical changes in definitions and calculation logic; Visual tag configuration interface supports drag-and-drop rule definition.

9. The CDP-based data fusion system according to claim 6, characterized in that: The privacy protection module also includes: Hybrid access control system based on RBAC and ABAC; Dynamic access policy engine that considers multiple factors such as time and location; Automated compliance audit tools to generate privacy impact assessment reports; The continuous data protection and rapid recovery module also includes: Point-in-time recovery function, with data rollback accurate to seconds; Business scenario recovery function, based on semantic integrity recovery; Sandbox test environment, verify the recovery effect before going online.

Images