Remote Snappable Linking for Consistent Incremental Cluster Backups
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data management systems face challenges in efficiently managing and recovering data across multiple clusters during failover and failback operations, leading to issues such as full backups and replication, inconsistent snapshot histories, and difficulties in identifying active members for backup and replication jobs.
Innovation Solution
Implementing a system for remote data object linking that creates a single entity for managing multiple remote data objects across different node clusters, designates a single active member for scheduling jobs, and performs incremental backups and replications, resolving discrepancies in active member designations and snapshot histories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If failover is performed to a secondary datacenter with new virtual machines, then data protection is improved, but snapshot history is lost causing full backups to be required
Solution Approach 1:
The patent merges the snapshot history from the primary datacenter with the new virtual machines in the secondary datacenter by creating a unified snapshot namespace. This allows the new VMs to inherit and continue using existing snapshot histories from the failed-over primary VMs, eliminating the need for full backups after failover.
Solution Approach 2:
The system copies snapshot metadata and history information from the primary datacenter to the secondary datacenter during the failover process. This copying of snapshot information enables the new VMs in the secondary datacenter to access and utilize the same snapshot history as the original VMs, maintaining incremental backup capabilities.
2Adaptability or versatility
If multiple remote data objects are managed separately across clusters, then cluster independence is maintained, but data management complexity increases
Solution Approach 1:
The patent implements a universal data object group structure that can contain data objects from multiple different clusters while maintaining a single unified management interface. The data object group acts as a multi-functional container that works across cluster boundaries, allowing administrators to manage remote data objects uniformly regardless of their physical cluster location.
Solution Approach 2:
The system introduces a data object group as an intermediary layer between individual data objects and the management interface. This intermediary abstracts the complexity of cross-cluster management by providing a unified view and control mechanism, while the underlying cluster independence is preserved through the group's ability to span multiple clusters.
3Productivity
If incremental backups are performed across clusters, then backup efficiency is improved, but consistent active member designation becomes difficult
Solution Approach 1:
The patent implements a feedback mechanism where the system continuously monitors and tracks the active member status of data objects across clusters. When a data object becomes inactive or is restored, the system automatically detects this change and reassigns the active member designation to an appropriate candidate, ensuring consistent tracking without manual intervention.
Solution Approach 2:
The system provides self-service functionality for active member management by automatically selecting and designating active members based on predefined criteria. When incremental backups are performed across clusters, the system autonomously manages the active member designations, reducing the operational burden on administrators while maintaining backup efficiency.
Data Source
AI summary
In some examples, a cluster comprises peer nodes and a distributed data store implemented across the peer nodes, a method of remote linking of data objects for data transfer between a first node cluster and a second node cluster among the peer nodes, the method comprising: creating a data object group including multiple remote data objects, wherein a plurality of remote data objects in the data object group represent a same first virtual machine and are registrable on at least the first and second node clusters of the peer DMS nodes; creating or identifying remote links to a plurality of the remote data objects in the data object group; designating a member of the data object group as an active member of the group; and assigning a task to the active member to be completed using remote links.