RAID Storage Parity Assignment for Parallel Write Access
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In traditional RAID storage systems, the need for a node to lock an entire stripe during writing operations to prevent access conflicts results in performance issues, especially when only partial data is being written, as this restricts parallel access and efficiency.
Innovation Solution
The solution involves assigning multiple parity values to different nodes within a stripe, allowing for parallel access and updating of data and parity values without locking other data spaces, thereby improving storage system efficiency by enabling nodes to execute read/write requests concurrently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a node locks the entire stripe during write operations to prevent access conflicts, then data consistency is ensured, but RAID performance deteriorates due to restricted parallel access
Solution Approach 1:
The patent segments the stripe into multiple data spaces, each associated with a specific node. Instead of locking the entire stripe, only the specific data space being written is locked. This segmentation allows other nodes to access different data spaces simultaneously, resolving the contradiction between data consistency (maintained through selective locking) and RAID performance (improved through parallel access).
Solution Approach 2:
The patent applies local quality by making each data space have distinct access control characteristics. Each data space within the stripe has its own locking scope, allowing fine-grained control where only the specific location being modified is locked, while other locations remain accessible. This localised locking approach maintains data consistency at the affected location while preserving overall system performance.
2Reliability
If a node locks the entire stripe during partial write operations, then access conflicts are prevented, but storage system efficiency deteriorates due to inability to execute concurrent operations
Solution Approach 1:
The stripe is segmented into multiple independent data spaces, each with its own access control. During partial write operations, only the specific data space being written is locked rather than the entire stripe. This enables other nodes to perform concurrent read/write operations on different data spaces, preventing access conflicts at the specific location while maintaining high storage system efficiency through parallelism.
Solution Approach 2:
The patent applies partial action by locking only the necessary portion (specific data space) rather than the entire stripe. This minimal locking approach provides sufficient conflict prevention for the operation at hand while leaving the rest of the stripe accessible, thereby maintaining storage system efficiency during partial write operations.
3Reliability
If the traditional locking mechanism is used for stripe access, then data integrity is maintained, but response time increases due to serialization of operations
Solution Approach 1:
By segmenting the stripe into multiple data spaces with independent locking, the patent enables parallel access to different segments. This reduces the serialization of operations and decreases response time, while data integrity is maintained through proper locking of the specific data space being accessed or modified.
Solution Approach 2:
The locking mechanism becomes dynamic and adaptive, locking only the specific data space needed for the current operation rather than statically locking the entire stripe. This dynamic approach reduces unnecessary waiting time for other operations while maintaining data integrity through context-aware locking.
Data Source
AI summary
A storage management technique involves: receiving from a requesting node a write request for writing target data into a first target storage space in a redundant array of independent disks (RAID); if a storage device associated with the first target storage space does not fail, acquiring first data stored in the first target storage space and a first parity value corresponding to the requesting node and stored in a parity storage space; determining a target parity value based on the target data, the first data, and the first parity value; and updating a stripe with the target data and the target parity value. Accordingly, locks caused by updating the parity value can be avoided, so that different nodes can perform parallel write to different storage spaces in the same stripe in the RAID.