Storage Slice Relocation Write I/O Handling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional storage systems face inefficiencies in slice relocation processes, leading to increased system resource consumption, lock contentions, and reduced performance due to unnecessary data copy operations during write I/O requests, especially when slices are being relocated.
Innovation Solution
The technique tracks the current data copy point within a slice being relocated, allowing write I/O requests to be copied only to the slice if the requested location is ahead of the copy point, thereby reducing redundant data copies and improving concurrency by mirroring only the portion of the slice that has been relocated.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data is copied to both the slice and destination slice for all write I/O requests during relocation, then data consistency is maintained, but system resource consumption increases and performance decreases
Solution Approach 1:
The slice relocation process is segmented into different regions: a first region that has been copied to the destination slice and a second region that has not been copied yet. Write I/O requests are handled differently based on which region they target, allowing the system to segment the consistency guarantee approach and avoid unnecessary copies.
Solution Approach 2:
Different quality of data consistency is applied locally to different regions of the slice. For the first region (already copied), full consistency is maintained by writing to both slices. For the second region (not yet copied), consistency is achieved through the relocation process itself, avoiding redundant write operations.
2Reliability
If all write I/O requests are copied to both slice and destination slice during relocation, then data integrity is preserved, but lock contentions increase and concurrency decreases
Solution Approach 1:
The slice is divided into a first region (copied) and second region (not copied). This segmentation allows the system to apply different locking strategies: the first region requires coordinated locking for consistency, while the second region can be accessed with simpler locking through the relocation process, reducing overall lock contentions.
Solution Approach 2:
The destination slice acts as an intermediary for the second region during relocation. Write requests to the second region are handled through the relocation mechanism rather than requiring direct coordination between source and destination, reducing lock contentions and improving concurrency.
3Reliability
If redundant data copy operations are performed during slice relocation, then data consistency is ensured, but backend I/O load increases and overall system performance decreases
Solution Approach 1:
Instead of performing complete data copying for all write requests, the system performs partial action only where necessary (first region). For the second region, the relocation process itself suffices, avoiding excessive copy operations and reducing backend I/O load while maintaining adequate consistency.
Solution Approach 2:
The patent extracts the unnecessary redundant copy operations from the write I/O handling process. By identifying and removing the requirement to copy to both slices for all regions, the system eliminates wasteful backend I/O operations while preserving essential data consistency through the relocation mechanism.
Data Source
AI summary
A method is used in managing data relocation in storage systems. A request is received to write to a slice of storage tier in a data storage system where the slice is being relocated. The relocation of the slice includes copying data of the slice to another slice in the data storage system. Location within the slice at which the request to write to the slice has been received is determined. Based on the determination, relocation of the slice of the storage tier is effected.