Data Migration to Non-Unified Storage Without Page Locks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing data migration process in hybrid memory systems is complex and inefficient due to the need for page locks in different storage media, leading to increased duration and reduced efficiency.
Innovation Solution
A data migration method that migrates data from a storage unit in a first storage medium to a second storage medium that does not participate in unified addressing, eliminating the need for locks and simplifying the migration procedure by decoupling address placement, thus reducing migration duration and improving efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the processor uses the page lock mechanism to migrate data between storage media, then data consistency is maintained, but the migration procedure becomes complex and duration increases
Solution Approach 1:
The patent extracts the locking mechanism from the data migration process by migrating data to storage units that do not participate in unified addressing. This eliminates the need to obtain and release page locks for both source and destination storage units, thereby removing the complexity of coordinate locking while maintaining data consistency through the exclusive access nature of non-unified addressing storage units.
Solution Approach 2:
The patent introduces an intermediary approach by using storage units in a second storage medium that do not participate in unified addressing as the destination for data migration. These intermediary storage units act as a buffer that receives data without requiring lock coordination, simplifying the migration procedure while ensuring data integrity through controlled access.
2Reliability
If the processor obtains and releases page locks for both old and new pages during migration, then data integrity is ensured, but migration duration increases
Solution Approach 1:
The patent removes the locking operations from the migration process by directing data to storage units that do not participate in unified addressing. This eliminates the time-consuming steps of obtaining locks before migration and releasing locks after migration, significantly reducing migration duration while maintaining data integrity through the exclusive access characteristics of the destination storage units.
3Ease of operation
If the processor migrates data to a storage unit that participates in unified addressing, then address mapping is simplified, but write-data consistency problems occur
Solution Approach 1:
The patent inverts the conventional approach by migrating data to storage units that do not participate in unified addressing, rather than to units that do. This inversion eliminates write-data consistency problems because the destination storage units are not accessible through the unified addressing mechanism, preventing concurrent write operations and ensuring data consistency without requiring complex locking protocols.
Data Source
AI summary
A data migration method includes, when a data migration request for a storage unit in a first storage medium is obtained, data in the storage unit is migrated to a storage unit that is in a second storage medium and does not participate in unified addressing. In a migration process, a data access request of an application is routed to the first storage medium or the second storage medium by querying a mapping relationship in a second unit mapping table and a migration state in a migration table. After the migration is completed, an address of a second storage unit is allocated to the application.