NDMP Daemon Proxying for Cluster Backup Volume Relocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In network clusters, using a network management protocol like NDMP for backup operations is problematic due to the unified namespace feature, which can result in errors when volumes are relocated, leading to wasted storage space and increased bandwidth and resources, as the backup manager may treat a relocated volume as a new volume, requiring unnecessary baseline backups.
Innovation Solution
A system and method that enables cluster-wide management of backup operations by extending the NDMP daemon with a processing module and a backup module, which determines the location of volumes using a volume location database and proxies backup operations between nodes if necessary, ensuring that backup operations are directed to the correct node for processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a unified namespace is implemented in a network cluster to allow any node to service any volume, then ease of management and volume mobility are improved, but errors occur when backup operations are directed to nodes that no longer service the target volume, leading to wasted storage space and increased bandwidth consumption
Solution Approach 1:
The patent introduces a volume location database as an intermediary component that maintains the mapping between volumes and the nodes that service them. Before executing backup operations, the system queries this database to determine the current location of the target volume, ensuring that backup commands are always directed to the correct node even after volume relocation. This intermediary mechanism resolves the contradiction by maintaining backup operation accuracy while preserving the unified namespace's ease of management.
2Adaptability or versatility
If volumes are relocated between nodes for load balancing or failover, then system flexibility and resource utilization are improved, but backup managers treat relocated volumes as new volumes requiring unnecessary baseline backups
Solution Approach 1:
The patent implements a preliminary action by querying the volume location database before initiating backup operations. This pre-check determines whether a volume has been relocated and allows the system to retrieve existing backup data from the new location, thereby avoiding unnecessary baseline backups. By performing this preliminary verification, the system maintains volume mobility while preventing waste of storage space.
3Productivity
If backup operations are performed without verifying volume location, then operational speed is improved, but errors occur and unnecessary baseline backups are triggered
Solution Approach 1:
The patent applies partial action by implementing a selective verification process. Rather than performing comprehensive checks for every backup operation, the system queries the volume location database only when necessary (e.g., when volume relocation is detected or at scheduled intervals). This approach maintains high backup operation speed while preventing the time loss associated with unnecessary baseline backups by applying verification selectively rather than universally.
Data Source
AI summary
A system and method enables cluster-wide management of backup operations issued by one or more backup managers to nodes of a cluster using a network management protocol, such as NDMP. An NDMP daemon executing on each node is extended to include a processing module and a backup module. The processing module interfaces with a backup manager to interpret the backup operation and, if necessary, forward (proxy) that operation to another NDMP daemon on another node of the cluster for processing in accordance with cluster's unified namespace.