Cache Page Reference Management for Storage Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage management systems face high input/output overhead due to the conventional one-to-one correspondence between cache pages and cache page references, which becomes inefficient when handling a large number of small read/write operations.
Innovation Solution
The system allows multiple cache page references to correspond to a single cache page, enabling the creation of virtual cache page references with the same pointer value, thereby reducing the need for data copying and improving IO performance by allowing parallel operations without allocating new cache pages.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If one cache page reference is allocated per cache page, then data management is simplified, but input/output overhead increases significantly
Solution Approach 1:
Multiple cache page references are merged to point to the same physical cache page, allowing several references to share a single underlying data structure. This reduces the total number of cache pages that need to be managed while maintaining the ability to perform multiple operations simultaneously through different references.
Solution Approach 2:
A single cache page serves multiple functions by being referenced by multiple cache page references. Each reference can independently initiate operations on the shared cache page, enabling one physical page to support multiple parallel operations rather than requiring separate pages for each operation.
2Productivity
If multiple cache page references are created for parallel operations, then IO performance improves, but data copying requirements increase
Solution Approach 1:
Multiple cache page references share the same physical cache page, eliminating the need to copy data between separate cache pages. Operations can proceed in parallel across different references without duplicating data, as all references access the same underlying memory region.
Solution Approach 2:
Instead of copying data between multiple cache pages, the system creates multiple references to the same cache page. This reference copying approach allows parallel operations without the data copying overhead that would occur if each reference had its own separate page.
3Reliability
If new cache pages are allocated for each operation, then operation independence is ensured, but storage overhead increases
Solution Approach 1:
Multiple operations share the same physical cache page through different references, reducing the total storage required. The system maintains operation independence through the reference layer, where each reference can be manipulated independently even though they point to the same underlying data.
Solution Approach 2:
The cache page reference acts as an intermediary between the operation and the physical cache page. This reference layer provides operation independence and addressing flexibility without requiring separate physical pages for each operation, thereby reducing storage overhead.
Data Source
AI summary
Techniques perform storage management. Such techniques involve, in response to an operation to be performed on data in a cache page, determining a first cache page reference corresponding to the cache page, the first cache page reference comprising a pointer value indicating the cache page. Such techniques further involve creating, based on the first cache page reference and the operation, a second cache page reference corresponding to the cache page, the second cache page reference comprising the pointer value. Such techniques further involve performing the operation on the data in the cache page via the second cache page reference. One cache page can correspond to a plurality of cache page references. Additionally, copy of data from one cache page to a further cache page can be effectively avoided, so as to enhance input/output performance and utilization rate of storage space.