Dual-Level RAID for SSDs Resolving Integrity and Space Trade-offs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing RAID systems, such as RAID 4 and RAID 5, face inefficiencies in managing data redundancy and integrity, especially when dealing with many open isolation units and frequent asynchronous updates, which can lead to increased storage costs and reduced performance.
Innovation Solution
Implementing a dual-level RAID scheme that combines RAID 1 for fine data granularity and RAID 4 or RAID 5 for high storage space utilization, allowing for efficient data mirroring and integrity data generation across multiple memory devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If RAID 4 or RAID 5 is applied to meet UBER requirement, then data integrity is improved, but storage space utilization deteriorates due to dedicated parity bits and frequent asynchronous updates requiring many open isolation units
Solution Approach 1:
The patent segments the RAID implementation into two distinct levels: Level 1 uses RAID 1 (mirroring) for fine-grained data protection at the page level, while Level 2 uses RAID 4 or RAID 5 (distributed parity) for batch-level integrity protection. This segmentation allows each level to operate optimally within its own granularity, resolving the contradiction between data integrity and storage space utilization by avoiding the need for dedicated parity bits at the fine-grained level.
Solution Approach 2:
The patent introduces a hierarchical dimension to RAID implementation by operating at two different granularity levels simultaneously. The first level operates at page level with mirroring, while the second level operates at batch level with distributed parity. This dimensional change allows the system to achieve both high storage space utilization and strong data integrity guarantees without the limitations of single-level RAID approaches.
2Reliability
If RAID 1 is applied for fine data granularity, then data protection capability is improved, but storage space consumption deteriorates due to data duplication
Solution Approach 1:
The patent segments the protection mechanism by applying RAID 1 only at the page level within batches, while using RAID 4/5 at the batch level. This segmentation allows fine-grained data protection without requiring complete data duplication across entire batches, as the distributed parity at the batch level provides complementary protection that reduces the overall storage overhead compared to pure RAID 1.
3Productivity
If many open isolation units are maintained for frequent asynchronous updates, then write performance is improved, but storage management complexity deteriorates
Solution Approach 1:
The patent implements periodic batch processing of pages within isolation units. Instead of managing each page update individually across many open isolation units, the system periodically closes batches when they reach a threshold, generates integrity data for the closed batches, and transitions to the next batch. This periodic action simplifies storage management by limiting the number of simultaneously open isolation units while maintaining high write performance through efficient batch processing.
Data Source
AI summary
This application is directed to storing data by distributing redundant data blocks across multiple memory devices (e.g., solid-state drives (SSDs)) and based on multiple RAID schemes. An electronic system mirrors user data on two distinct memory devices. Integrity data is generated based on the user data and stored on an integrity memory device. In accordance with a determination that the integrity data of the user data is stored on the integrity memory device, the electronic system releases the user data mirrored on at least one of the two distinct memory devices. In some embodiments, a copy of the user data is also stored on one of a plurality of data memory devices. The integrity data is generated based on the user data and stored on the integrity memory device, in accordance with a determination that the plurality of data memory devices is filled.