Cluster Backup Querying for VM Location and HA Status
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for cluster backups and restores are inefficient due to the inability to identify properties of virtual resources, leading to unnecessary data duplication, inaccurate backup scheduling, and offline resource issues during HA VM restores, as they are unaware of VM locations and HA status across cluster nodes.
Innovation Solution
A computer system that queries both virtual and cluster subsystems across nodes to aggregate information on VMs, their locations, and HA status, enabling accurate scheduling and operation of backup and restore processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the virtual subsystem performs backup operations without querying cluster subsystem information, then the backup process is simple and fast, but the backup accuracy and reliability deteriorate due to inability to identify HA VM status and location
Solution Approach 1:
The patent introduces an intermediary mechanism where the virtual subsystem queries the cluster subsystem to obtain HA VM status and location information. This intermediary information exchange enables the virtual subsystem to make informed backup decisions without directly managing cluster resources, thus maintaining backup speed while improving accuracy and reliability.
Solution Approach 2:
The patent implements a feedback loop where the virtual subsystem receives status information from the cluster subsystem about HA VMs and uses this feedback to adjust its backup operations. This feedback mechanism ensures that backup operations are performed on the correct VMs at the correct locations, improving reliability without significantly impacting backup speed.
2Reliability
If the system performs volume-level backup of entire cluster shared volume, then all data is backed up ensuring completeness, but data redundancy increases and storage efficiency decreases
Solution Approach 1:
The patent extracts only the necessary data for backup by querying the cluster subsystem to identify which specific VMs and data portions need to be backed up. Instead of backing up the entire cluster shared volume, the system extracts and backs up only the relevant data portions, thereby maintaining data completeness while eliminating redundant backups and improving storage efficiency.
3Reliability
If cluster resources are taken offline before restoring HA VM, then the restore operation can proceed safely, but service availability decreases and downtime increases
Solution Approach 1:
The patent applies preliminary action by querying the cluster subsystem beforehand to identify the current location and status of HA VMs. This advance information allows the restore operation to proceed without taking cluster resources offline, as the system already knows where to restore the VM and can coordinate the restore process to maintain service availability while ensuring restore safety.
4Device complexity
If the virtual subsystem operates without querying other cluster nodes, then the system complexity is reduced, but the ability to track VM migration and location deteriorates
Solution Approach 1:
The patent segments the information gathering process by having each cluster node query its own virtual subsystem for local VM information and then aggregate this information through the cluster subsystem. This segmentation approach maintains relatively simple individual node operations while collectively achieving complete VM location tracking across the cluster, thus balancing system complexity with information completeness.
Data Source
AI summary
A system and method for identifying properties of virtual resources to efficiently perform backups and restores of cluster data. A cluster of nodes is coupled to a data storage medium. A node receives a request for a backup or a restore of cluster data. In response to this request, the node queries a cluster subsystem and a virtual subsystem of all other cluster nodes for identification of VMs, a subset of corresponding stored data, and an identification of VMs which are highly available (HA). In response to receiving query responses, the node aggregates the results and sends them to a backup server. These aggregated results may then be used to schedule subsequent backup and restore operations. In addition, the node may use the results to complete the current backup or restore operation.