Data Copy Avoidance Across Storage Nodes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data copy avoidance protocols are inefficient when implemented across multiple physical storage nodes, as they require data to be transmitted twice between nodes, leading to resource wastage and performance bottlenecks.
Innovation Solution
A method where a first storage node sends a data access request to a second storage node, with the second node allocating a storage space and providing the address of a corresponding storage space in its mirrored cache, allowing the first node to perform data access directly without mirroring back to the initiating node.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If current DCA protocol is used across storage nodes, then data access efficiency is improved within a single node, but data transmission overhead increases when operations are redirected to another storage node
Solution Approach 1:
The patent introduces a buffer area at the destination storage node as an intermediary to receive and hold data locally. This buffer acts as a mediator between the source and destination nodes, allowing the initiating node to access data from its local cache without requiring the destination node to mirror data back, thus eliminating redundant transmissions while maintaining DCA benefits
Solution Approach 2:
The patent segments the data access process into distinct phases: data transmission from source to destination node, local caching at the destination node, and direct access by the initiating node. This segmentation allows each phase to be optimized independently, with the buffer area enabling the initiating node to access data locally without involving the destination node in the access phase
2Ease of operation
If data is mirrored back to the initiating storage node in current DCA solution, then data access can proceed, but the operation requires twice data transmission between nodes which deteriorates efficiency
Solution Approach 1:
The patent performs preliminary action by pre-transmitting data from the source node to the destination node's buffer area before the initiating node needs to access it. This preliminary transmission allows the data to be staged locally at the destination node, eliminating the need for subsequent mirroring operations and enabling direct access without redundant transmissions
Solution Approach 2:
The patent inverts the traditional mirroring approach by having the destination node retain data in its buffer area for direct access, rather than mirroring it back to the source node. This inversion changes the data flow direction: instead of source→destination→source, the system uses source→destination→initiating node directly, eliminating the return transmission
Data Source
AI summary
Embodiments of the present disclosure relate to methods and apparatuses for data copy avoidance where after a data access request is received from the first storage node, what is sent by a second storage node to the first storage node is not an address of a second storage space in a second mirrored cache, but an address of a first storage space in a first cache corresponding to the second storage space. In this way, data access may be implemented directly in the first cache on the first storage node, and can reduce data communication across different storage nodes, eliminate potential system performance bottlenecks, and enhance data access performance.