Database Disaster Recovery Using Modified-Page Bitmap Sync

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The existing database disaster recovery systems face challenges with high network resource consumption and prolonged recovery time (RTO) due to large data transmission requirements during WAL log-based stream replication for data synchronization between primary and disaster recovery clusters.

Innovation Solution

Implementing a page modification information bitmap to identify and synchronize only modified pages between clusters, reducing data transmission and utilizing standby node devices for direct storage without log replay, thereby optimizing data synchronization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If WAL log-based stream replication is used for data synchronization, then data consistency between primary and disaster recovery clusters is maintained, but network resource consumption increases and recovery time is prolonged

Engineering Contradiction:
Improvedata consistencyVSAvoidnetwork resource consumption
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The patent extracts only the modified pages from the complete data set using a bitmap mechanism. The bitmap tracks which pages have been modified, allowing the system to send only those specific pages to the disaster recovery cluster instead of all WAL logs, thereby reducing network resource consumption while maintaining data consistency.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent segments the data into individual pages and uses a bitmap to identify which segments (pages) need to be synchronized. This segmentation approach allows selective transmission of only modified pages rather than transmitting all data or all WAL logs, reducing network overhead.

Inventive Principle:
Principle #1Segmentation

2Reliability

If WAL log-based stream replication is used for data synchronization, then data consistency between primary and disaster recovery clusters is maintained, but recovery time objective (RTO) is prolonged

Engineering Contradiction:
Improvedata consistencyVSAvoidrecovery time objective
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent performs preliminary actions by maintaining a bitmap that tracks modified pages in advance. When a disaster occurs, the system can quickly identify and transmit only the modified pages using the pre-maintained bitmap information, rather than replaying all WAL logs from scratch, thus significantly reducing RTO while ensuring data consistency.

Inventive Principle:
Principle #10Preliminary action

3Reliability

If all data is transmitted during synchronization, then complete data backup is ensured, but network resources are heavily occupied

Engineering Contradiction:
Improvedata backup completenessVSAvoidnetwork resource occupation
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The patent extracts only the necessary modified pages using the bitmap mechanism, sending only those pages to the disaster recovery cluster. This extraction approach ensures that all modified data is backed up completely while avoiding transmission of unchanged data, thereby reducing network resource occupation.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP4660811A1Data disaster recovery method, apparatus and system, node device, and standby node device
Publication Date: 2025.12.10 HUAWEI CLOUD COMPUTING TECHNOLOGIES CO LTD
  • EP4660811A1 patent drawingFigure 1~2
  • EP4660811A1 patent drawingFigure 3~4
  • EP4660811A1 patent drawingFigure 5

AI summary

This application discloses a data disaster recovery method, apparatus, and system, a node device, and a standby node device, and relates to the field of data storage. A database disaster recovery system includes a primary cluster and a disaster recovery cluster. Data synchronization is periodically performed between the primary cluster and the disaster recovery cluster. In a data synchronization process, a node device in the primary cluster may determine, based on a page modification information bitmap, a to-be-synchronized page on which data modification is performed in the node device, replicate data of the to-be-synchronized page, and send a page address of the to-be-synchronized page and the data of the to-be-synchronized page to a standby node device corresponding to the node device, where the standby node device is a device in the disaster recovery cluster. The node device in the primary cluster synchronizes only data of a page on which data modification is performed to the standby node device in the disaster recovery cluster, so that an amount of data transmitted between the primary cluster and the disaster recovery cluster can be reduced, network resources occupied for data transmission can be reduced, and an RTO can be effectively shortened.