Fault Tolerant Server Memory Synchronization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The availability of fault-tolerant servers is decreased due to the time required for memory copy operations between online and offline subsystems, which causes temporary cessation of information processing and reduces system availability.
Innovation Solution
A fault-tolerant server configuration that includes an entire copy unit for background copying of data from the online subsystem to the offline subsystem without stopping information processing, along with data detection and change data copy units to identify and synchronize changed data, ensuring continuous operation during memory duplication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If memory copy is performed from the online system to the offline system before synchronizing, then the memory content is synchronized, but the information processing stops during the copy process reducing availability
Solution Approach 1:
The patent performs preliminary memory copy actions by copying only the necessary initial pages (e.g., kernel space pages) before full system synchronization. This preliminary action enables the offline system to start processing earlier while the remaining memory pages are copied in the background, thus improving availability without compromising eventual memory synchronization completeness.
2Reliability
If all memory pages are copied collectively, then complete memory synchronization is achieved, but the copy time increases causing longer system downtime
Solution Approach 1:
The patent segments the memory copy process into multiple phases: first copying critical pages (kernel space, essential system data), then copying remaining pages in the background. This segmentation allows the system to achieve functional synchronization with critical data while non-critical data copying continues without blocking system operations, thereby reducing overall system downtime.
Solution Approach 2:
The patent performs preliminary copy of essential memory pages before full synchronization is complete. By identifying and copying only the most critical pages first (such as kernel space pages that contain essential system operations), the system can become operational sooner while the remaining memory pages are transferred in the background, significantly reducing the perceived downtime.
3Manufacturing precision
If the offline system is kept stopped to ensure memory consistency, then synchronization accuracy is improved, but the fault tolerant capability is reduced
Solution Approach 1:
The patent performs preliminary synchronization of critical memory pages before the offline system needs to take over. By ensuring that essential pages (kernel space, critical data structures) are copied and synchronized in advance, the offline system can be started and begin processing sooner while maintaining sufficient consistency for fault tolerance operations.
Data Source
AI summary
A fault tolerant server according to the present invention configured to duplicate information processing by an online subsystem and an offline subsystem, the fault tolerant server operates to: execute entire copy processing for copying all data being stored in the memory of the online subsystem into the memory of the offline subsystem without stopping execution of information processing by the processor of the online subsystem, before start of duplication; detect data, the data satisfying a criterion indicating that content of data is changed during the entire copy processing, among data being stored in the memory of the online subsystem; and copy the detected data from the memory of the online subsystem into the memory of the offline subsystem.