Unbroken Snapshot Sequence Computation for High-Availability Clusters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multi-tiered disaster recovery architectures, restoring a desired data state from snapshots distributed across nodes is challenging due to broken snapshot chains, requiring full backups that are time-consuming and resource-intensive.
Innovation Solution
The implementation of techniques that identify and compute unbroken snapshot sequences across multiple nodes, optimizing memory usage, CPU cycles, and network bandwidth by determining difference calculations to combine snapshots and reduce data transfer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If full backup is used to restore data state from broken snapshot chains, then data restoration reliability is improved, but restoration time and network I/O requirements increase significantly
Solution Approach 1:
The patent segments the snapshot restoration process by identifying and utilizing unbroken snapshot chains across multiple nodes. Instead of requiring a complete full backup, the system segments the restoration into two parts: (1) retrieving the most recent unbroken snapshot from available nodes, and (2) applying incremental changes from subsequent snapshots. This segmentation enables faster restoration by avoiding the need to re-transfer entire datasets.
Solution Approach 2:
The patent performs preliminary actions by pre-identifying and pre-assembling unbroken snapshot sequences across multiple nodes before actual restoration is needed. The system continuously monitors and maintains knowledge of which snapshots form unbroken chains, so when restoration is required, the process can immediately utilize these pre-identified sequences without needing to scan through all available snapshots.
2Reliability
If full backup is used to restore data state from broken snapshot chains, then data restoration reliability is improved, but network I/O requirements increase significantly
Solution Approach 1:
The patent extracts only the essential information needed for restoration by identifying unbroken snapshot chains and retrieving only those specific snapshots rather than transferring complete backup images. The system extracts the minimal set of data required to reconstruct the desired state, significantly reducing network I/O requirements compared to full backup approaches.
Solution Approach 2:
The patent utilizes copying by creating and maintaining multiple copies of snapshots across different nodes in a distributed manner. Each node stores a subset of snapshots, and the system tracks which copies form unbroken chains. During restoration, only the necessary copied snapshots are retrieved and combined, rather than requiring complete data duplication from a single source.
3Measurement precision
If multi-tiered storage architecture is implemented with frequent snapshots, then recovery freshness is improved, but storage complexity and data management difficulty increase
Solution Approach 1:
The patent implements feedback mechanisms where nodes continuously report their snapshot inventories and chain statuses to a coordinating system. This feedback loop enables the system to automatically identify unbroken chains, track recovery freshness, and manage the multi-tiered architecture dynamically without manual intervention, reducing the operational complexity despite the increased structural complexity.
Solution Approach 2:
The patent creates universality by designing a unified snapshot management approach that works across all tiers of the storage architecture. The same principles for identifying unbroken chains and retrieving minimal data apply consistently from primary storage through backup sites, simplifying data management despite the hierarchical structure. The system treats all storage locations uniformly in terms of their contribution to forming valid restoration chains.
Data Source
AI summary
Methods, systems and computer program products for high-availability computing. In a computing configuration comprising a primary node, a first backup node, and a second backup node, a particular data state is restored to the primary node from a backup snapshot at the second backup node. Firstly, a snapshot coverage gap is identified between a primary node snapshot at the primary node and the backup snapshot at the second backup node. Next, intervening snapshots at the first backup node that fills the snapshot coverage gap are identified and located. Having both the backup snapshot from the second backup node and the intervening snapshots from the first backup node, the particular data state at the primary node is restored by performing differencing operations between the primary node snapshot, the backup snapshot from the second backup node, and the intervening snapshots of the first backup node.