Disk Array Write Method Using Persistent Memory Cache
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In RAID systems, incomplete write operations due to power failures can lead to data inconsistency, rendering failed data unrecoverable and causing the 'write hole' problem, which is typically addressed by reserving journal areas but at the cost of reduced performance due to extra disk operations.
Innovation Solution
The method involves determining if an incomplete write operation would render data unrecoverable, recovering the failed data, and storing it in a persistent memory before writing it to the target storage area, thereby avoiding data loss and improving performance by eliminating the need for extra disk operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If journal area is reserved to prevent write hole, then data reliability is improved, but write performance deteriorates due to extra disk I/O operations
Solution Approach 1:
The patent introduces a cache memory as an intermediary layer between the host system and the disk array. The cache temporarily stores write data and metadata, allowing the system to acknowledge write completion to the host before the data is physically written to disk. This intermediary buffer eliminates the need for journal areas while maintaining data reliability through cached persistence.
Solution Approach 2:
The patent replaces the mechanical disk I/O operations (journal area writes) with in-memory caching operations. By using volatile cache memory for temporary storage and metadata tracking, the system substitutes slow mechanical disk access with fast electronic memory access, thereby improving write performance while still ensuring data reliability through cache management mechanisms.
2Reliability
If journal area is used to restore consistency, then data integrity is improved, but processing speed deteriorates due to additional write operations
Solution Approach 1:
The patent performs preliminary actions by pre-allocating cache memory spaces and pre-establishing metadata structures before actual write operations occur. The cache is prepared in advance to hold both data and consistency information, eliminating the need for post-write consistency restoration operations that would slow down processing.
Solution Approach 2:
The patent merges the data storage function and the consistency tracking function into a single cache memory structure. Instead of separate journal areas for consistency information, the cache integrates both data and metadata (including consistency markers and recovery information), reducing the number of I/O operations and improving processing speed while maintaining data integrity.
3Reliability
If extra disk operations are performed for write protection, then data safety is improved, but system efficiency deteriorates
Solution Approach 1:
The cache acts as an intermediary that handles all write protection and safety operations in memory, preventing the need for corresponding disk operations. The cache manages data safety through checksums, metadata tracking, and persistence protocols entirely in volatile memory, eliminating redundant disk I/O and improving system efficiency while maintaining data safety.
Data Source
AI summary
Techniques involve writing to a disk array. In response to a request for writing target data to a target storage area in a stripe of the disk array including at least one failed storage area, it is determined whether incompletion of the writing of the target data is to cause failed data in the at least one failed storage area to be unrecoverable. In response to determining that the incompletion of the writing is to cause failed data to be unrecoverable, the failed data is recovered. The recovered data is stored in a persistent memory, and the target data is written in the target storage area. In this way, the efficiency and performance of writing processing of the disk array having a failed disk is increased significantly.