Data Fusion Across Data Centers for Consistent Service Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In remote multi-active architectures, data synchronization delays between data centers can lead to inconsistent data, causing errors and data omissions during service switching, which current methods fail to address effectively, compromising data integrity and continuity.
Innovation Solution
A data fusion method that identifies and updates data based on transaction statuses across multiple data centers, ensuring accurate sequencing and integrity by comparing transaction statuses rather than just update times, using unique identifiers and transaction status machines to align and update data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data synchronization is implemented between multiple data centers, then data availability and service continuity are improved, but data consistency and accuracy deteriorate due to synchronization delays
Solution Approach 1:
The patent applies preliminary action by establishing a data fusion mechanism that proactively compares and reconciles data from multiple data centers before service switching occurs. The system pre-establishes data correspondence relationships and transaction status mappings between data centers, enabling accurate data alignment during failure scenarios without waiting for synchronization to complete naturally.
Solution Approach 2:
The patent implements feedback through the data fusion process that continuously compares transaction statuses and data versions across data centers. The system uses feedback loops to identify inconsistencies, determine correct data versions based on transaction status mappings, and resolve conflicts, thereby maintaining data consistency while enabling service continuity.
2Reliability
If data is copied to multiple data centers for backup, then disaster recovery capability is improved, but data accuracy deteriorates due to synchronization delays
Solution Approach 1:
The patent uses copying by creating replicated data structures in multiple data centers, but enhances this with intelligent data fusion that compares copied data versions. Instead of simple replication, the system copies data with associated transaction status information and uses this metadata to accurately reconstruct the correct data state during recovery operations.
Solution Approach 2:
The patent applies parameter changes by transforming the data representation to include transaction status fields and version information alongside the actual data. This parameter enhancement enables the system to distinguish between different versions of copied data and accurately determine which version should be restored, thereby maintaining data accuracy during disaster recovery.
3Ease of manufacture
If update time is used to determine data correctness, then implementation simplicity is improved, but data integrity deteriorates due to omissions and inconsistencies
Solution Approach 1:
The patent introduces an intermediary mechanism in the form of transaction status mapping that mediates between data versions from different data centers. Instead of directly comparing data based on update times, the system uses transaction status information as an intermediary to determine the correct data version, thereby maintaining data integrity while keeping the implementation relatively simple through automated status comparison.
Data Source
AI summary
A data fusion method and apparatus, and a device and a storage medium are provided. The method includes: acquiring, from a first data center, a plurality of pieces of first data to be compared that are within a preset time period, and acquiring, from a second data center, a plurality of pieces of second data to be compared that are within the preset time period; then, determining the same data unique identifier in the plurality of pieces of first data to be compared and the plurality of pieces of second data to be compared, and taking the same data unique identifier as a first data identifier; and acquiring first data to be compared that corresponds to the first data identifier, and acquiring second data to be compared that corresponds to the first data identifier.