Distributed all-flash storage data writing method and device and medium
By multiplexing the EC data blocks in the erasure group as replica cache data blocks, the replica management cost and write amplification problems in traditional solutions are solved, and efficient data writing and storage performance improvements are achieved.
CN119739340AActive Publication Date: 2025-04-01LANGCHAO ELECTRONIC INFORMATION IND CO LTD
5 Cites 2 Cited by
Patent Information
- Application Number
- CN202412000403.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-31
- Publication Date
- 2025-04-01
- Estimated Expiration
- 2044-12-31
AI Technical Summary
Technical Problem
The write performance optimization solution of traditional distributed all-flash storage systems in erasure scenarios leads to additional replica management costs and write amplification problems.
Method used
By multiplexing the EC data blocks in the erasure group as the replica cache data blocks, small block IO aggregate writes are performed, and erasure code calculations are performed when the preset threshold is reached to reduce the cost of repeated writes and replica management.
Benefits of technology
It reduces the amount of data written in the background, reduces the repeated writes of erased data, avoids additional replica management costs, and improves the storage performance of distributed all-flash storage systems.
✦ Generated by Eureka AI based on patent content.
Abstract
The invention discloses a distributed all-flash storage data writing method and device and a medium, relates to the technical field of distributed storage, and is used for optimizing all-flash storage writing performance. The invention provides a distributed all-flash storage data writing method aiming at the problem that the optimization of the writing performance of the current aggregation writing scheme has a further improvement space. According to the method, an EC data block is used as a copy data block in a multiplexing manner. In this way, the purpose of optimizing writing of aggregating small IO blocks into large IO blocks for writing can still be achieved, a copy cache partition does not need to be additionally established, and therefore additional copy management cost is avoided. Moreover, according to the method, when the service data is persistently stored in the EC form, the repeated writing of the erasure data is reduced, so that the data writing overhead is reduced, and the storage performance of the distributed all-flash storage system is improved.
Need to check novelty before this filing date? Find Prior Art