Mirror Storage Resynchronization Latency Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In storage virtualization systems, particularly in switch-based environments, mirror storage element resynchronization during data volume recovery introduces significant latency due to the need for transitions between I/O and control modules, causing delays and inaccessibility until recovery is complete.
Innovation Solution
A method that processes read operations by determining the state of a region using a synchronization map, allowing the operation to be handled by either the I/O module or control module based on this state, reducing the number of faults and latency during resynchronization by distributing processing across multiple modules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If mirror storage element resynchronization is performed using traditional fault-based module transitions, then data consistency is restored, but significant latency and downtime occur
Solution Approach 1:
The patent pre-positions resynchronization data in the I/O module's buffer memory before faults occur. When a fault happens, the control module can immediately retrieve and apply the pre-prepared resynchronization data without waiting for data transfer, thus eliminating the time-consuming data transfer step that normally causes latency.
Solution Approach 2:
The patent introduces an intermediary buffer memory in the I/O module that acts as a mediator between the storage elements and the control module. This buffer temporarily holds resynchronization data, allowing the I/O module to handle resynchronization operations independently without constant control module intervention, thereby reducing latency.
2Ease of operation
If read operations are processed during resynchronization, then access availability is maintained, but data inconsistency may occur
Solution Approach 1:
The patent applies different quality levels to different data regions by using a synchronization map. The map identifies which regions are synchronized and which are not, allowing the system to permit read operations in synchronized regions while blocking or redirecting reads in unsynchronized regions, thus maintaining availability where safe while ensuring consistency where needed.
Solution Approach 2:
The patent implements a feedback mechanism through the synchronization map that continuously tracks the consistency state of different data regions. This feedback information is used by the I/O module to dynamically adjust read operation handling, allowing reads in consistent regions while preventing reads in inconsistent regions until resynchronization completes.
Data Source
AI summary
A method and system for mirror storage element resynchronization in a storage virtualization device is disclosed. According to one embodiment of the present invention, a read operation on a region of a data volume is received at a first module of a first storage virtualization device where the data volume is associated with at least a first mirror storage element and a second mirror storage element, a state indicated by a bit corresponding to the region of the data volume within a first synchronization map is determined in response to receiving the read operation, the read operation is processed using the first module if the bit is determined to indicate a first state and using a second module of the first storage virtualization device if the bit is determined to indicate a second state.