Hybrid Data Replication Bypassing Secondary Storage Wait
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data replication technologies incur significant overhead due to the primary storage device waiting for notification from the secondary storage device, especially over long distances, which affects transaction processing performance and availability.
Innovation Solution
Implementing a hybrid data replication mechanism where the host determines the type of data to write, sending separate write commands to primary and secondary storage devices in parallel, allowing the primary storage device to notify the host independently of the secondary storage device's completion status, thereby bypassing the need for the primary to wait for secondary confirmation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the primary storage device waits for notification from the secondary storage device before notifying the host, then data replication reliability is improved, but transaction processing time increases significantly
Solution Approach 1:
The system performs preliminary actions by having the primary storage device notify the host of write completion before the secondary storage device finishes writing. This allows the host to proceed with subsequent operations without waiting for remote replication completion, reducing transaction processing time while maintaining data integrity through alternative mechanisms.
Solution Approach 2:
The invention segments the notification process into independent events: the primary storage device notifies the host of local write completion independently, while separately tracking remote replication status. This segmentation allows the host to proceed without waiting for remote replication, resolving the time-delay contradiction.
2Reliability
If synchronous replication is used to ensure data consistency, then data integrity is improved, but system performance deteriorates due to waiting overhead
Solution Approach 1:
The system dynamically adjusts replication behavior based on data type classification. For critical data types, synchronous replication ensures integrity; for less critical data, asynchronous or delayed replication is used, allowing the system to balance integrity requirements with performance demands dynamically.
Solution Approach 2:
The invention changes the replication parameter from strictly synchronous to a flexible mode where the primary device can notify the host before remote completion. This parameter change allows the system to maintain data integrity through logging and recovery mechanisms while significantly improving system performance.
3Reliability
If data is replicated over long distances, then availability and redundancy are improved, but notification overhead increases by 10 μs per kilometer
Solution Approach 1:
The invention extracts the notification function from the replication completion dependency. The primary storage device notifies the host of write completion independently of remote replication status, removing the causal link between remote notification and host notification. This eliminates the distance-dependent overhead from the critical transaction path.
Data Source
AI summary
A system for accelerating database transaction processing by controlling data replication is provided. The system includes a first control unit configured to manage a first storage device and at least one second control unit configured to manage a second storage device. The first control unit writes first data to the first storage device and sends the first data to the second control unit in response to receiving from a host a first write command including the first data. The first control unit writes second data to the first storage device without sending the second data to the second control unit in response to receiving from the host a second write command. The second control unit writes the first data to the second storage device in response to receiving the first data. The second control unit writes the second data to the second storage device in response to receiving from the host a third write command.