N-site Asynchronous Replication Cycle Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data replication systems, such as synchronous and asynchronous RDF systems, face challenges in balancing latency and data consistency across storage devices, particularly when there is a long distance between primary and secondary storage devices, leading to increased latency and potential data loss in case of failures.
Innovation Solution
A computing system with recovery groups that include both synchronous and asynchronous data centers, where cycle switching for asynchronous data transfer is coordinated by the source data center, allowing for efficient data chunking and transmission to ensure low latency and data consistency across multiple recovery groups.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If synchronous RDF transfer is used, then data consistency between primary and secondary storage devices is ensured, but latency increases due to waiting for acknowledgment
Solution Approach 1:
The system performs preliminary actions by pre-positioning data in buffers at the secondary storage device before actual write operations complete. This allows the acknowledgment to be sent immediately when data arrives at the buffer, rather than waiting for the full write cycle to complete, thereby reducing latency while maintaining data consistency through buffered confirmation.
2Loss of time
If asynchronous RDF transfer is used, then latency is reduced, but data consistency and recoverability are compromised
Solution Approach 1:
The system implements feedback mechanisms through sequence numbers and cycle tracking that allow the primary storage device to monitor and verify asynchronous transfers. By tracking which data blocks have been successfully transferred and acknowledged, the system can ensure data consistency and enable recovery even in asynchronous mode, as the primary device can identify and retransmit any missing or inconsistent data.
Solution Approach 2:
The system prepares recovery information and transfer state data in advance during normal operation. This preliminary preparation of transfer metadata and sequence information enables rapid recovery operations without requiring full data re-synchronization, thus maintaining both low latency and data consistency through pre-computed recovery paths.
3Device complexity
If multiple recovery groups are managed independently, then system complexity is reduced, but resource utilization and efficiency decrease
Solution Approach 1:
The system merges the management of multiple recovery groups under a unified control architecture at the primary storage device. This centralized coordination allows the primary device to manage data distribution across multiple secondary devices and recovery groups simultaneously, optimizing resource utilization and enabling efficient load balancing while maintaining manageable complexity through standardized protocols and interfaces.
Data Source
AI summary
A computing system includes at least one source data center that generates data and a plurality of recovery groups, coupled to the at least one source data center, where each of the recovery groups includes a synchronous data center that receives synchronous data from the at least one source data center and includes an asynchronous data center that receives asynchronous data from the at least one source data center. Cycle switching for asynchronous data transfer is coordinated by the at least one source data center for a first subset of recovery groups containing at least two of the recovery groups. The computing system may also include a second subset of the recovery groups, different from the first subset, where cycle switching for asynchronous data transfer is coordinated by the at least one source data center for the second subset.