Logical Partitioning for High-Speed Storage Access and Data Protection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current storage systems face challenges in fully utilizing the performance of high-speed devices like SSDs due to bottlenecks in storage controllers and incomplete data protection, especially when high-speed devices are directly coupled to servers, which either limits access performance or compromises data protection.
Innovation Solution
Logical partitioning of computers into compute and storage partitions, where I/O operations are sorted and routed accordingly to optimize access to storage device groups, allowing for flexible use of high-speed access and data protection as needed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a storage controller is used to provide data protection functions, then data protection is improved, but the I/O performance of high-speed devices such as SSD cannot be fully brought out due to bottleneck
Solution Approach 1:
The system is segmented into two logical partitions: a compute partition that handles I/O requests and a storage partition that handles storage control functions. This segmentation allows I/O requests to be processed in the compute partition without being bottlenecked by the storage controller, while data protection functions are maintained in the storage partition.
Solution Approach 2:
A communication interface acts as an intermediary between the compute partition and storage partition. The compute partition sends I/O requests through this interface to access storage devices directly, bypassing the storage controller bottleneck while the storage partition maintains data protection functions.
2Productivity
If high-speed devices are directly coupled to server, then I/O performance is improved, but storage functions such as data protection function provided in storage controller cannot be fully utilized
Solution Approach 1:
The system separates storage functions into a dedicated storage partition that can provide data protection functions, while the compute partition directly accesses high-speed storage devices. This allows both direct access performance and storage functions to coexist.
Solution Approach 2:
The storage partition serves multiple functions: it provides data protection functions, manages storage devices, and communicates with the compute partition. This multi-functionality allows the system to utilize both direct access performance and storage controller functions.
3Speed
If write-through method is used for cache application, then Read performance is improved, but Write performance cannot be fully brought out and data protection is not necessary for all write data
Solution Approach 1:
The system dynamically determines the access path for each I/O request based on its characteristics. For read requests, data can be retrieved from cache for high speed, while for write requests, the system can choose to bypass the storage controller for performance or use it for data protection, making the system adaptive to different workloads.
Data Source
AI summary
In a computer, a logical partition for calculation in which an OS and an application operate and a logical partition for storage for providing a storage function are constructed. In the logical partition for calculation, a device corresponding to a storage device is provided, while the logical partition for storage provides a volume. A memory space that can be shared by the both logical partitions is prepared, and management information describing a sorting destination or a sorting method of an I/O request issued by an application is provided in the memory. If the logical partition for calculation receives an I/O request from the application, the partition refers to the management information and sorts the I/O request to the storage device or the logical partition for storage. The logical partition for storage processes the received I/O request by the storage function and transmits the result to the storage device.