Stripe Defragmentation Based on Access Frequency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Striped RAID storage volumes in distributed storage systems face performance degradation due to fragmentation and the intensive IO load associated with defragmentation and parity recomputation, especially in shared storage systems where resources are shared.
Innovation Solution
The implementation of a method that maintains stripe metadata with timestamps and read counts to determine read access frequency, allowing for targeted defragmentation and rebuild processes based on the temperature index of each stripe, thereby optimizing IO operations and reducing system load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If stripe defragmentation is performed to improve storage performance, then IO overhead is reduced, but system performance degrades due to intensive IO load
Solution Approach 1:
The patent applies local quality by differentiating treatment of stripes based on their access frequency characteristics. Frequently accessed stripes (hot stripes) are prioritized for defragmentation and rebuild operations, while less frequently accessed stripes are deferred. This selective approach optimizes system performance by focusing resources on the most impactful stripes rather than uniformly processing all stripes.
Solution Approach 2:
The patent implements preliminary action by maintaining stripe metadata that includes access frequency information in advance. This metadata is continuously updated and used to pre-identify which stripes should be defragmented or rebuilt first, allowing the system to proactively manage fragmentation before it severely impacts performance, rather than reactively responding to performance degradation.
2Productivity
If defragmentation is performed on parity-based RAID storage volumes, then storage performance is improved, but system performance degrades due to recomputing parity stripes and metadata updates
Solution Approach 1:
The patent applies local quality by selectively applying defragmentation and parity recomputation only to stripes that exhibit fragmentation and high access frequency. This avoids the overhead of processing all stripes uniformly, reducing the time lost to parity recomputation and metadata updates while still achieving the necessary storage performance improvements.
Solution Approach 2:
The patent implements partial action by performing defragmentation and rebuild operations on only a subset of stripes - specifically those identified as hot stripes with high access frequency. This partial approach is sufficient to improve overall system performance without incurring the full overhead of processing all stripes, thereby reducing the time loss associated with parity recomputation and metadata updates.
3Productivity
If full stripe writes are performed to achieve lower IO overhead, then write performance is improved, but free capacity becomes fragmented making full stripe writes difficult
Solution Approach 1:
The patent implements preliminary action by proactively performing defragmentation of frequently accessed stripes before they become highly fragmented. By maintaining these stripes in an optimized state with sufficient free capacity, the system ensures that full stripe writes can continue to be performed efficiently without being hindered by fragmentation, thus preserving write performance over time.
Data Source
AI summary
A method comprises maintaining stripe metadata corresponding to stripes of data stored on a plurality of storage devices. The stripe metadata comprises a first timestamp, a second timestamp and a read count for each stripe. The method comprises obtaining a read request associated with a given stripe and having an associated timestamp and determining whether or not a time period for the given stripe has been exceeded. The method comprises updating the stripe metadata corresponding to the given stripe based at least in part on the determination and the associated timestamp and determining a read access frequency of the given stripe based at least in part on the read count, the first timestamp and the second timestamp of the updated stripe metadata. The method comprises performing at least one of a defragmentation process and a rebuild process based at least in part on the determined read access frequency.