Data Reconstruction Using Split Check Matrix Submatrices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing data reconstruction methods in storage systems require a large number of read requests and data transmissions, which occupy significant network bandwidth and affect performance, especially when reconstructing lost blocks in a stripe across multiple storage devices.
Innovation Solution
A data reconstruction method that determines a lost block in a stripe and reconstructs it using results obtained from multiple storage devices, reducing the need for reading unaffected blocks and minimizing data transmission by splitting the check matrix into submatrices and processing these results to reconstruct the block directly, thereby reducing CPU overhead and network bandwidth usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional block reconstruction method is used (reading all unlost blocks and transmitting them to control device), then block reconstruction can be completed, but network bandwidth consumption increases significantly
Solution Approach 1:
The patent divides the check matrix into multiple submatrices, where each submatrix corresponds to a specific storage device. This segmentation allows the system to process and transmit only the necessary portions of data related to each device, rather than transmitting the entire check matrix and all unlost blocks, thereby reducing network bandwidth consumption while maintaining reconstruction reliability
Solution Approach 2:
The patent extracts only the essential information needed for reconstruction by calculating and transmitting only the relevant portions of check blocks corresponding to lost blocks, rather than transmitting all unlost blocks. This extraction approach reduces the volume of data transmitted over the network while still providing sufficient information for accurate block reconstruction
2Reliability
If traditional block reconstruction method is used (sending N read requests to read unlost blocks), then block reconstruction can be completed, but CPU overhead increases
Solution Approach 1:
The patent performs preliminary calculation of check block values based on submatrices before the actual reconstruction process. By pre-calculating and preparing the necessary check information in a structured manner, the system reduces the computational burden during reconstruction, lowering CPU overhead while ensuring reliable block recovery
Solution Approach 2:
The patent introduces submatrices as intermediary structures that mediate between the check matrix and the reconstruction process. These submatrices pre-organize the check information in a way that simplifies the reconstruction calculation, reducing the complexity of CPU operations required while maintaining reconstruction accuracy
3Reliability
If traditional block reconstruction method is used (transmitting N blocks and reconstructed block), then block reconstruction can be completed, but reconstruction time increases
Solution Approach 1:
The patent segments the check matrix into device-specific submatrices, which enables parallel processing of reconstruction operations for different storage devices. This segmentation allows multiple calculations to occur simultaneously, reducing the overall reconstruction time while maintaining the reliability of block recovery through the distributed submatrix approach
Data Source
AI summary
A method describes that a lost first block in a first stripe is directly reconstructed based on a first result obtained from a target quantity of storage devices, so that there is no need to read an unlost block in the first stripe. The first block may be reconstructed provided that the target quantity of first results are obtained. An amount of data of the first result is less than that of the unlost block in the first stripe. Therefore, a data transmission process occupies less network bandwidth, and this improves block reconstruction performance.