Cache Data Migration via Management Information Transfer
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
During job migration in a business system, the inefficiency in data access occurs due to the time-consuming synchronization of cache data and storage device data, especially when the difference between the two is large, leading to delayed job migration and reduced data access efficiency for migrated jobs.
Innovation Solution
A method where a first control processor creates and transmits management information indicating the storage location and update status of cache data to a second control processor, allowing the second processor to acquire and store the cache data directly, thereby skipping the synchronization process and enabling direct migration of cache data to the migration destination node.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the file system synchronizes cache data and storage device data before job migration, then data consistency is maintained, but migration time increases significantly when the difference between cache data and storage data is large
Solution Approach 1:
The system performs preliminary actions by creating management information that records the storage locations and update statuses of cache data blocks before migration. This allows the destination node to directly retrieve only the necessary cache data blocks without performing full synchronization, thus maintaining data consistency while reducing migration time.
Solution Approach 2:
The invention extracts only the necessary cache data blocks that need to be migrated by using management information to identify which blocks have been updated. Instead of synchronizing all cache data, the system extracts and transfers only the specific blocks that require migration, significantly reducing the time and resources needed.
2Reliability
If the file system synchronizes all cache data with storage device data, then complete data consistency is achieved, but data access efficiency decreases due to the time-consuming synchronization process
Solution Approach 1:
The system performs partial synchronization by using management information to identify and migrate only the cache data blocks that have been updated. This partial action approach maintains necessary data consistency while avoiding the overhead of synchronizing all cache data, thus preserving data access efficiency.
Solution Approach 2:
The invention creates management information as a copy or representation of the cache data status, which includes storage locations and update flags. This management information serves as a guide for selective data migration without requiring physical copying or synchronization of all cache data, maintaining efficiency while ensuring consistency.
3Reliability
If the migrated job accesses the storage device again for previously cached data, then data accuracy is ensured, but data access efficiency is reduced
Solution Approach 1:
The system performs preliminary migration of cache data blocks to the destination node before the job resumes execution. By using management information to identify which blocks need migration, the system ensures that the migrated job can access previously cached data directly from the destination cache without needing to re-access the storage device, thus maintaining both data accuracy and access efficiency.
Data Source
AI summary
A data migration method includes creating, by a first control processor that controls a first cache memory storing first cache data cached from first storage data stored in a storage, first management information including information indicating a storage location of the first cache data on the first cache memory and information indicating whether or not the first storage data has been updated in accordance with an update of the first cache data for each block of a predetermined data size in the first cache memory, when a program that accesses the first cache data migrates to a different node, transmitting, by the first control processor, the first management information to a second control processor that controls a second cache memory capable of being accessed by the program after migration to the different node.