Rolling Node Snapshots for Consistent Storage Mirroring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current storage networks face inefficiencies when mirroring data across multiple nodes, as they require quiescing each node to create a snapshot, leading to temporary unavailability and performance issues during data replication for disaster recovery and high-availability purposes.
Innovation Solution
The method involves obtaining snapshots of individual nodes in a rolling or asynchronous manner within a clustered storage network, allowing for minimal disruption to other nodes, and compiling these snapshots to create an eventually consistent view of the entire system, which can be used for mirroring without halting system operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If nodes are quiesced to create a synchronized snapshot of the cluster, then data consistency for mirroring is improved, but node availability and network performance deteriorate
Solution Approach 1:
The patent divides the cluster snapshot operation into individual node snapshots. Instead of quiescing the entire cluster simultaneously, each node is snapshot independently while remaining operational. This segmentation allows data consistency to be achieved through compilation of individual snapshots without sacrificing node availability.
Solution Approach 2:
The system performs preliminary actions by having each node continuously maintain snapshot capability and using event-driven triggers to initiate snapshots only when needed. This allows the system to prepare snapshot data in advance without requiring full cluster quiescence, thereby maintaining productivity while ensuring data consistency when mirroring is required.
2Manufacturing precision
If nodes are quiesced to establish a snapshot, then replication accuracy is improved, but network throughput and user access deteriorate
Solution Approach 1:
The patent enables continuous useful action by allowing nodes to remain operational and serve client requests while snapshot operations proceed independently. The compilation of individual node snapshots creates an eventually consistent view that maintains replication accuracy without interrupting network throughput or user access to stored data.
Solution Approach 2:
The system introduces an intermediary compilation process that aggregates individual node snapshots into a unified cluster snapshot. This intermediary layer ensures replication accuracy by maintaining consistency across nodes without requiring the nodes themselves to be quiesced, thereby preserving network throughput and user access.
3Reliability
If all nodes are suspended for clustering snapshot, then data integrity for disaster recovery is improved, but system operational efficiency deteriorates
Solution Approach 1:
The patent implements dynamic snapshot compilation where the system adapts to node availability and operational states. Individual node snapshots are taken and compiled dynamically without requiring all nodes to be suspended simultaneously. This dynamic approach maintains data integrity for disaster recovery while preserving operational efficiency by allowing nodes to remain active.
Solution Approach 2:
By segmenting the snapshot process into individual node operations rather than a monolithic cluster-wide quiescence, the system achieves data integrity through compiled snapshots while maintaining operational efficiency. Each node can be snapshot independently and remain operational, eliminating the need to suspend entire systems for disaster recovery preparation.
Data Source
AI summary
Systems and methods which provide for obtaining snapshots of one or more nodes, while having minimal impact on performance and throughput of the storage network are provided. Embodiments may involve a method of obtaining snapshots of individual node data within a clustered storage network having a plurality of nodes without requiring quiescing of every node. The snapshots may be obtained in a rolling fashion, e.g., moving across the individual nodes of a cluster either sequentially or non-sequentially. In other embodiments snapshots may be obtained in an asynchronous manner. The obtained snapshots then collectively define an eventually consistent view of the entire system. Such a view may be utilized to provide for mirroring functionality, e.g. to provide for DR and/or HA capabilities.