Cluster Backup Query Mechanism for Duplicate Data Elimination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In clustered computing environments, the lack of fast live migration of applications and virtual machines (VMs) due to dependency on disk resources leads to inefficiencies, and duplicate backups occur during volume-level and agent-based backup operations, as nodes have no knowledge of data usage by other nodes, resulting in redundant data copies.
Innovation Solution
Implementing a cluster shared volume accessible to all nodes, utilizing the Volume Shadow Copy Service (VSS) to identify and exclude data subsets used by applications on other nodes from the initial volume-level backup, ensuring that only unique data is backed up, thereby avoiding duplicate copies during subsequent agent-based backups.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a volume-level backup operation is performed on a shared volume in a cluster, then the entire volume is backed up, but data used by applications on other nodes is backed up multiple times (duplicate backups)
Solution Approach 1:
Before performing the volume-level backup, the system performs preliminary queries to all cluster nodes to identify which data portions will be backed up by subsequent agent-based backups. This preliminary action enables the volume-level backup to exclude those portions, preventing duplicate backups while maintaining comprehensive data protection.
Solution Approach 2:
The system implements a feedback mechanism where each cluster node responds to queries about their data usage patterns. This feedback information is used by the backup system to intelligently determine which data portions should be excluded from the volume-level backup, thereby avoiding duplicates while ensuring all necessary data is protected.
2Ease of manufacture
If agent-based backup temporarily disables applications to backup individual files, then file-level control is achieved, but application unavailability increases
Solution Approach 1:
The backup process is segmented into two distinct phases: a volume-level backup phase that captures data without disabling applications, and an agent-based backup phase that handles specific file-level requirements. This segmentation allows the system to achieve file-level control when needed while minimizing overall application disruption.
Solution Approach 2:
The system performs preliminary identification of which data portions need agent-based backup processing before actually disabling applications. This allows for better planning and coordination, enabling the system to minimize application unavailability time while still achieving the necessary file-level backup control.
3Stability of the object's composition
If nodes depend on disk resources for VM mobility, then storage stability is maintained, but fast live migration of VMs becomes impossible
Solution Approach 1:
The system uses copying mechanisms to enable VM migration. Instead of moving physical disk resources, the system creates copies of data portions that can be quickly accessed by different nodes. This copying approach maintains storage stability while enabling fast live migration by allowing the destination node to access data through these copies without requiring physical disk movement.
Data Source
AI summary
A system and method for avoiding duplicate backups of data in a volume backup image. A cluster of nodes is coupled to a data storage medium configured to store data corresponding to a shared volume. A node receives a request for the backup of the shared volume. In response to this request, the node queries all other nodes in the cluster for identification of a subset of the shared volume data that should be excluded from the backup. The identified subset of the data may correspond to at least a file under control of a database application or other particular application. It may be known that this identified subset of the data will be backed up by a subsequent agent-based backup operation after the volume backup. In response to receiving all query responses, the node initiates a backup corresponding to the data in the shared volume excluding the identified subset of the data.