Middle Map Architecture for vSAN Snapshot I/O Overhead Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing virtual storage area network (vSAN) architecture faces significant I/O overhead during segment cleaning and snapshot management due to the need to update multiple snapshot logical maps when physical data blocks are dynamically relocated, leading to poor snapshot performance and increased I/O operations.
Innovation Solution
A two-layer snapshot extent mapping architecture is introduced, where logical block addresses (LBAs) are mapped to middle block addresses (MBAs), and the MBA is then mapped to physical block addresses (PBAs), allowing only the MBA's PBA to be updated, reducing the need for multiple updates across snapshot logical maps and minimizing I/O overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If physical data blocks are dynamically relocated during segment cleaning, then storage efficiency is improved, but I/O overhead increases due to multiple snapshot logical map updates
Solution Approach 1:
The patent introduces a middle map as an intermediary layer between the logical block address (LBA) and physical block address (PBA). When physical blocks are relocated during segment cleaning, only the middle map needs to be updated to reflect the new PBA mapping, while the snapshot logical maps remain unchanged. This intermediary layer decouples the snapshot management from physical block relocation operations, eliminating the need for multiple metadata write operations across different snapshot logical maps.
2Reliability
If multiple snapshot logical maps are updated when physical blocks are relocated, then snapshot consistency is maintained, but snapshot performance deteriorates
Solution Approach 1:
The middle map serves as a mediator that absorbs the impact of physical block relocation. Snapshot logical maps continue to reference MBAs without needing updates, while the middle map dynamically translates these MBAs to current PBAs. This maintains snapshot consistency by preserving the original LBA-to-MBA mappings while handling physical relocation transparently through the middle map layer.
Solution Approach 2:
The mapping architecture is segmented into multiple layers: LBA to MBA mapping in snapshot logical maps, and MBA to PBA mapping in the middle map. This segmentation allows independent management of snapshot references and physical block locations, enabling physical relocation without propagating changes to snapshot logical maps.
3Device complexity
If direct LBA to PBA mapping is used, then mapping simplicity is maintained, but metadata update overhead increases during block relocation
Solution Approach 1:
The middle map acts as an intermediary layer that adds minimal complexity to the mapping structure while significantly reducing metadata update overhead. Instead of directly mapping LBA to PBA and requiring updates to all snapshot logical maps during relocation, the system maps LBA to MBA (in snapshot logical maps) and MBA to PBA (in middle map), isolating update operations to a single middle map structure.
Data Source
AI summary
A method for block addressing is provided. The method includes moving content of a data block referenced by a logical block address (LBA) from a first physical block corresponding to a first physical block address (PBA) to a second physical block corresponding to a second PBA, wherein prior to the moving a logical map maps the LBA to a middle block address (MBA) and a middle map maps the MBA to the first PBA and in response to the moving, updating the middle map to map the MBA to the second PBA instead of the first PBA.