RAID Metadata Recovery via Binary Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional RAID systems are inflexible in adding new storage capacity, requiring increments of multiple drives at once, which becomes impractical as individual drive capacities increase, and previous methods for recovering the last allocated logical block address for metadata storage are time-consuming and resource-intensive, especially when copies become corrupted.
Innovation Solution
The implementation of mapped RAID technology, which allows individual storage drives to be added to a RAID protected system and uses a binary search across the last two RAID extents to quickly recover the last allocated logical block address within a logical address space, ensuring seamless operation of host I/O requests.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional RAID systems are used for storage, then data redundancy and protection are provided, but the system becomes inflexible in adding new storage capacity requiring increments of multiple drives at once
Solution Approach 1:
The patent segments the RAID storage system into individual drive components that can be independently managed. Each drive maintains its own metadata structures and allocation information, allowing individual drives to be added or removed without requiring changes to the entire RAID array configuration. This segmentation enables flexible capacity expansion while maintaining RAID protection through independent drive management.
2Reliability
If traditional methods are used to recover the last allocated logical block address, then the system can restore metadata storage continuity, but the recovery process becomes time-consuming and resource-intensive
Solution Approach 1:
The patent implements preliminary action by maintaining persistent storage of the last allocated logical block address in metadata structures on each drive. Before potential failure scenarios occur, the system continuously updates and preserves this allocation state information in non-volatile storage. When recovery is needed, this pre-preserved information allows immediate restoration of metadata storage continuity without time-consuming searches or scans.
3Reliability
If multiple copies of the last allocated logical block address are stored for safety, then reliability is improved, but the risk of all copies becoming corrupted simultaneously increases
Solution Approach 1:
The patent segments the storage of allocation state information across multiple independent drives rather than relying on multiple copies on a single drive. Each drive maintains its own metadata structures with allocation information specific to that drive. This segmentation isolates corruption risks to individual drives, so that failure or corruption on one drive does not affect the allocation state information on other drives, maintaining system reliability while reducing the risk of simultaneous corruption.
Data Source
AI summary
A logical unit of non-volatile data storage is provided for persistently storing metadata. The logical unit is provided by allocating RAID (Redundant Array of Independent Disks) extents to the logical unit, and allocating logical block addresses for the logical unit from within a logical address space of the logical unit to store blocks of metadata written to the logical unit. In response to detecting that a stored indication of a last allocated logical block address for the logical unit has been lost, two RAID extents that were last allocated to the logical unit are identified, and a binary search is performed across logical block addresses within the two RAID extents that were last allocated to the logical unit to locate the last allocated logical block address for the logical unit within the logical address space of the logical unit.