Triangular Asynchronous Replication Switchover
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data replication systems, such as synchronous and asynchronous RDF systems, face challenges with high latency and data consistency issues, especially when transferring data between distant storage devices, and existing solutions like SAR processes require significant overhead and may result in data loss during failures.
Innovation Solution
Implementing a triangular asynchronous replication system that swaps primary and backup groups between data centers, allowing for synchronous and asynchronous mirroring to ensure low latency and data consistency, with the ability to create and reverse data mirroring relationships and initiate multisession control to manage data synchronization and recovery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If synchronous RDF transfer is used to ensure data consistency between primary and secondary storage devices, then data reliability is improved, but write latency increases due to waiting for acknowledgment
Solution Approach 1:
The system segments the acknowledgment process by introducing intermediate acknowledgment points. Instead of waiting for a single end-to-end acknowledgment from the secondary storage device, the system breaks down the transfer into segments that can be acknowledged independently, allowing the host to proceed with subsequent operations while full synchronization completes in the background.
Solution Approach 2:
The system performs preliminary data buffering and pre-synchronization actions before the host needs to access the data. By pre-copying data to the secondary storage device and maintaining a buffer of synchronized data, the system ensures data consistency is already established before write operations complete, eliminating the need for the host to wait for synchronization acknowledgment.
2Speed
If semi-synchronous RDF transfer is used to reduce write latency by acknowledging writes immediately, then write speed is improved, but data consistency is compromised because previous transfers must be acknowledged first
Solution Approach 1:
The system maintains continuous data transfer operations between primary and secondary storage devices without interruption. By keeping the synchronization pipeline continuously active and using background threads to handle data copying, the system ensures that data consistency is maintained continuously rather than in discrete acknowledged steps, allowing immediate write acknowledgment while preserving reliability.
3Speed
If asynchronous background copy is used to reduce latency, then write speed is improved, but data consistency cannot be guaranteed between primary and secondary storage devices
Solution Approach 1:
The system implements feedback mechanisms where the secondary storage device continuously reports synchronization status and data consistency metrics back to the primary storage device. This feedback loop allows the system to monitor and maintain data consistency even during asynchronous operations, providing visibility into the replication state without blocking write operations.
4Reliability
If SAR process is used to handle failures in triangular replication, then system recovery capability is improved, but operational overhead increases significantly
Solution Approach 1:
The system implements self-service failure recovery mechanisms where storage devices automatically detect, diagnose, and recover from failures without requiring external SAR process intervention. By embedding fault tolerance and automatic recovery logic directly into the storage device firmware and replication protocol, the system maintains high reliability while eliminating the complex operational overhead of manual SAR processes.
Data Source
AI summary
A primary group may be swapped with a synchronous backup group where triangular asynchronous replication is being provided between the primary group, the synchronous backup group and an asynchronous backup group. Swapping may include halting work at the primary group, transferring pending mirrored data from the primary group to an asynchronous backup group, creating a data mirroring relationship between a storage volume at the synchronous backup group and a storage volume at the asynchronous backup group, reversing a data mirroring relationship between the storage volume at the primary group and the storage volume at the synchronous backup group, and resuming work at the asynchronous backup group.