Storage System Buffer Switching for Data Consistency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In storage systems with synchronous and asynchronous copy methods, maintaining data order and synchronizing buffer regions across multiple storage devices is challenging, especially when one device fails, leading to potential data inconsistencies and increased response time due to round-trip time (RTT) issues.
Innovation Solution
A storage system architecture with multiple control devices and buffers, where synchronous copy is performed between the first and second devices, and asynchronous copy is performed between the first and third devices, with a mechanism to switch buffer regions and maintain data order using delta copy sessions and buffer index tables to ensure data consistency across all devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If synchronous copy is performed from first storage device to second storage device, then data consistency is maintained, but round-trip time increases as distance between devices increases, deteriorating responsivity to writing
Solution Approach 1:
The patent segments the copy operation into two distinct modes: synchronous copy for the primary backup (first to second storage device) and asynchronous copy for the remote backup (first to third storage device). This segmentation allows the system to maintain data consistency where needed while avoiding RTT penalties in remote scenarios, as the asynchronous copy does not wait for acknowledgment from the distant third device.
Solution Approach 2:
The patent applies different copy qualities to different destinations based on their roles and distances. The second storage device (local/primary backup) receives synchronous copy with full consistency guarantees, while the third storage device (remote backup) receives asynchronous copy where eventual consistency is acceptable. This local quality differentiation resolves the contradiction by matching the copy method to the specific requirements of each destination.
2Reliability
If asynchronous copy is performed with order maintained using record buffers, then data order is preserved, but when first storage device fails, the second storage device cannot immediately take over I/O processing because data may not be identical with third storage device
Solution Approach 1:
The patent employs preliminary actions in two forms: (1) The second storage device continuously receives and stores copies of data written to the first storage device, maintaining a ready state for fail-over. (2) Buffer regions are pre-established and can be switched to maintain data order during asynchronous operations. These preliminary preparations ensure that when fail-over occurs, the second device can immediately take over I/O processing without waiting for data synchronization with the third device.
Solution Approach 2:
The patent introduces dynamic buffer region switching capability that allows the system to adapt between maintaining strict data order and enabling rapid fail-over. The buffer management mechanism can dynamically adjust which data portions require strict ordering versus which can be processed asynchronously, balancing data integrity with operational continuity during fail-over scenarios.
3Adaptability or versatility
If third storage device is located at long distance, then large scale disaster coverage is improved, but synchronizing data with first storage device deteriorates responsivity
Solution Approach 1:
The patent segments the backup strategy into hierarchical levels: the second storage device serves as the primary backup for rapid fail-over, while the third storage device serves as a remote disaster-recovery backup. This segmentation allows the system to place the third device at a geographically distant location for disaster coverage without impacting responsivity, since asynchronous copy is used for this remote destination and does not require real-time acknowledgment.
4Reliability
If buffer regions are switched to maintain data order, then data consistency is improved, but device complexity increases due to buffer management mechanisms
Solution Approach 1:
The patent introduces buffer regions as intermediary storage areas between the write operations and the actual data transfer to backup devices. These buffer regions act as mediators that temporarily hold data during buffer switches, ensuring data consistency without requiring complex real-time coordination during fail-over. The buffer management mechanism, while adding some complexity, provides a structured intermediary layer that simplifies the overall data flow control and consistency maintenance.
Data Source
AI summary
A first control unit of a first control device executes: a synchronous copy process for writing data from a host device into a first storage medium, transmitting the data to a second control device, and replying to the host device; a storage process for storing the data written in the first storage medium into a first region, instructing the second control device on a buffer switch when a predetermined condition is satisfied, and switching a storage destination to a second region; and an asynchronous copy process for transmitting the data in the first region and the second region to a third control device at a time, respectively. A second control unit stores the data transmitted from the first control unit in a second storage medium and a third region, and when the buffer switch is instructed, stores the data in the second storage medium and a fourth region.