Hot Migration Virtual Storage Device for Zero-Downtime Data Transfer
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data migration methods in high-availability systems require substantial downtime due to the need to shut down the system during data copying, leading to potential data corruption and prolonged unavailability, especially with large storage devices.
Innovation Solution
A method for 'hot migration' that creates a virtual storage device mapped to both the original and destination storage units, allowing the system to remain available for read and write operations, with write requests directed to the destination unit and read requests handled from either unit based on data validity, enabling concurrent data copying and minimizing downtime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the system is shut down to copy data during migration, then data integrity can be maintained, but system availability becomes zero and downtime increases to hours
Solution Approach 1:
The migration process is divided into distinct phases: pre-migration preparation, concurrent read-write operation phase, and post-migration switching. During the concurrent phase, read requests are served from the original storage device while write requests are directed to the new device, allowing data to be copied without shutting down the system.
Solution Approach 2:
The new storage device is prepared and mounted before the actual data migration begins. The system is configured to direct write requests to the new device in advance, so when the migration starts, the new device is already ready to receive data without requiring system shutdown.
2Productivity
If the system remains running during data copying, then system availability is maintained, but data corruption occurs due to concurrent writes
Solution Approach 1:
Different storage devices are assigned different functions: the original storage device handles read requests while the new storage device handles write requests. This differentiation of local qualities allows concurrent operations without data corruption, as each device serves its designated function exclusively.
Solution Approach 2:
A storage virtualization server acts as an intermediary between the system and the storage devices. It manages the mapping between logical volumes and physical devices, directing read requests to the original device and write requests to the new device, thereby preventing data corruption during concurrent operations.
3Adaptability or versatility
If conventional disk migration is used, then data can be migrated to replacement devices, but the process requires substantial downtime of many hours
Solution Approach 1:
The system dynamically switches the mapping between logical volumes and physical devices during the migration process. The storage virtualization server updates the mapping in real-time, allowing the system to transition from using the original storage device to using the new device without interruption and minimal downtime.
4Productivity
If RAID storage is used for migration, then some availability is maintained, but migration options are limited and it functions better as failure protection than migration solution
Solution Approach 1:
The storage virtualization server provides multi-functionality by handling both data migration and maintaining system availability simultaneously. It can direct reads to the original device and writes to the new device, making it a universal solution that works for various storage device types and migration scenarios without being limited to specific RAID configurations.
Data Source
AI summary
Methods and software are provided for “hot migration” of data an original storage unit for a target system to destination storage unit. In one embodiment, a virtual storage device is provided, having a control block mapped to both the original storage unit, as well as to the destination storage unit. The target system is stopped, the virtual storage device is substituted for the original storage unit, and the target system is restarted. The virtual storage device directs requests to the destination storage unit, and fulfills read requests from whichever of the two storage units has a valid copy of data requested (depending, e.g., on whether the data is as yet unchanged, has been re-written, or has been copied, during the migration process). Unchanged data blocks are copies from the original storage unit to the destination storage unit. The target system is stopped, the virtual storage device is disconnected from the target system, and the destination storage unit is connected in its place. Software may also be provided to automate these processes.


