Clustered Storage Takeover Without Partner Mode
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage system architectures require partner mode configurations for failover operations, leading to increased code complexity and operational bandwidth consumption, and do not allow for seamless takeover of disks without maintaining separate infrastructure for partner and local modes.
Innovation Solution
A system that enables a surviving node in a clustered storage system to assert ownership of disks from a failed node, creating newly assimilated aggregates without the need for partner mode, by changing SCSI reservations and RAID ownership, and ensuring data consistency through NVRAM replay and fsid duplication correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If partner mode configuration is used for failover operations, then storage system reliability is improved, but code complexity and operational bandwidth consumption increase
Solution Approach 1:
The patent extracts and eliminates the partner mode infrastructure from the failover system. Instead of maintaining separate partner and local modes with complex differentiation logic, the system uses a single unified mode where any surviving node can assume ownership of failed node's disks through SCSI reservation pre-emption and RAID group reassignment, thereby reducing code complexity while preserving failover reliability
Solution Approach 2:
The patent implements universality by making any surviving node in the cluster capable of performing takeover operations without requiring specific partner mode configuration. The failover mechanism becomes universal and applicable to any node pair, eliminating the need for specialized partner infrastructure and simplifying the overall system architecture
2Reliability
If partner mode configuration is used for failover operations, then storage system reliability is improved, but operational bandwidth consumption increases
Solution Approach 1:
The patent removes the partner mode-specific communication and coordination overhead from the system. By eliminating the need for separate partner infrastructure, the system reduces operational bandwidth consumption associated with partner mode maintenance while keeping the essential failover capability through direct SCSI reservation pre-emption and RAID group ownership transfer
3Reliability
If separate infrastructure for partner and local modes is maintained, then failover operations can be performed, but device complexity increases
Solution Approach 1:
The patent merges the previously separate partner mode and local mode infrastructures into a single unified failover mechanism. The system now uses the same RAID group ownership and SCSI reservation pre-emption logic for all failover operations, eliminating the need for separate infrastructure maintenance and reducing overall system complexity while preserving failover functionality
Solution Approach 2:
The patent creates a universal failover mechanism that works for all node failures without requiring mode-specific infrastructure. Any surviving node can take over any failed node's disks using the same procedures, making the infrastructure simpler and more maintainable while ensuring reliable failover operations
Data Source
AI summary
A system and method for performing a takeover procedure is disclosed. A plurality of storage system nodes is coupled in a cluster. One or more of the plurality of storage system nodes in the cluster is notified to takeover at least a portion of a writable storage device serviced by another storage system node in the cluster. In response to the notification, the portion of the writable storage device is assimilated into at least one of the notified storage system nodes.