Linked Clone Storage via UPIT B+ Tree Snapshots
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current snapshot technologies, particularly redo-log and tree-based methods, face scalability issues when managing large numbers of snapshots, with redo-log techniques requiring intensive computations and tree-based techniques having complex intertwined structures, making it difficult to efficiently manage and replicate storage objects.
Innovation Solution
The implementation of a universal point-in-time (UPIT) manager using a copy-on-write (COW) B+ tree structure to organize storage objects and their snapshots, allowing for efficient creation and management of snapshots within a single storage structure, and the use of linked clones to replicate storage objects by referencing existing data and metadata, reducing the need for complete data copying.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If redo-log based snapshot techniques are used, then snapshot changes can be maintained in separate redo logs, but the system cannot scale to manage a large number of snapshots and requires intensive computations to consolidate across different snapshots
Solution Approach 1:
The patent merges multiple snapshot data structures into a single unified B+ tree structure. Instead of maintaining separate redo logs for each snapshot, all snapshot data is organized in one consolidated B+ tree, eliminating the need for complex consolidation operations across multiple separate structures. This directly addresses the scalability and computational intensity issues of redo-log based approaches.
Solution Approach 2:
The patent uses copy-on-write technology to create snapshots by copying only the necessary data blocks to new locations in the B+ tree structure, rather than maintaining complete separate copies or complex redo logs. This enables efficient snapshot creation and management at scale by minimizing redundant data storage and access operations.
2Productivity
If tree-based snapshot techniques are used, then scalability to manage large numbers of snapshots is improved, but the snapshot structure becomes intertwined and complex
Solution Approach 1:
The patent segments the snapshot management structure into distinct, well-defined components within the B+ tree hierarchy, including snapshot roots, data blocks, and metadata sections. Each component has a specific function and clear boundaries, which simplifies navigation and management compared to intertwined structures. The segmentation is achieved through organized data blocks and clear tree hierarchy levels.
Solution Approach 2:
The patent creates a universal B+ tree structure that serves multiple functions: storing snapshot data, maintaining metadata, supporting replication, and enabling efficient navigation. This single unified structure replaces multiple specialized data structures, reducing overall system complexity while maintaining scalability. The B+ tree universally handles both data and metadata management for all snapshots.
3Reliability
If complete data copying is used to replicate storage objects, then replicated storage objects are created, but storage needs and computational resources are not reduced
Solution Approach 1:
The patent implements nested doll by creating linked clones where the cloned storage object nests within the parent object's B+ tree structure. The clone shares the same underlying data blocks as the parent, with only minimal metadata and difference tracking stored separately. This nested relationship enables replication with dramatically reduced storage requirements, as the clone essentially lives inside the parent's data structure.
Solution Approach 2:
The patent uses selective copying to create linked clones, copying only the essential metadata and data block references needed to establish the clone relationship, rather than copying complete data sets. The actual data blocks are shared between parent and clone through the B+ tree structure, minimizing storage requirements while maintaining replication reliability.
Data Source
AI summary
System and method for creating linked clones of storage objects stored in a storage uses a target storage structure that includes a linked clone being created from a selected snapshot of a storage object and surface snapshots. The linked clone is created in the target storage structure with references to the selected snapshot of the storage object in a source storage structure without completely copying the selected snapshot to the target storage structure. The surface snapshots, which correspond to other snapshots of the storage object in the source storage structure, are created in the target storage structure with references to the other snapshots without completely copying the other snapshots to the target storage structure to produce the linked clone with similar snapshots as the storage object at a point in time when the selected snapshot was captured.