Failover Cache Data Preservation via Converted Fast Write Destaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In data storage subsystems with a pair of clusters, when one cluster fails over to the local cluster, the dual mode DASD fast write data stored in volatile cache becomes the only copy, making it vulnerable, as it is not adequately protected and cannot be destaged to disk storage due to potential drive or adapter issues.
Innovation Solution
The data storage subsystem converts cache storage dual mode DFW data to converted fast write data, prioritizes it for destaging to disk storage, and allocates non-volatile storage tracks to ensure data safety by reconverting it back to DFW data, using a retro-store control to emulate a host adapter and manage track ID entries, ensuring data integrity during failover.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If dual mode DASD fast write data is stored in volatile cache storage for fast access, then data access speed is improved, but data reliability deteriorates because the data becomes vulnerable when it becomes the only copy after failover
Solution Approach 1:
The system performs preliminary actions by converting cache storage DFW data to converted fast write data and prioritizing it for destaging to disk storage before failover completes. This ensures data is moved from volatile to non-volatile storage proactively, preventing data loss while maintaining access performance during normal operation.
Solution Approach 2:
The patent introduces an intermediary mechanism (converted fast write data and priority destaging list) between the volatile cache and permanent disk storage. This intermediary allows the system to manage data transition and prioritize critical data movement, resolving the conflict between speed and reliability by creating a buffer zone for data protection.
2Reliability
If data is prioritized for destaging to disk storage, then data protection is improved, but system productivity deteriorates due to additional processing overhead
Solution Approach 1:
The system applies local quality by creating a specific priority destaging list only for converted fast write data that needs protection, rather than treating all data uniformly. This targeted approach protects critical data while minimizing impact on overall system productivity by focusing resources only where needed.
Solution Approach 2:
The patent implements partial action by prioritizing only the necessary converted fast write data for destaging, rather than processing all cache data with equal priority. This selective approach provides adequate protection for vulnerable data without the excessive processing overhead of treating all data equally, thus balancing reliability and productivity.
3Reliability
If non-volatile storage tracks are allocated for converted fast write data, then data integrity is improved, but device complexity increases due to track allocation and reconversion processes
Solution Approach 1:
The system employs self-service by having the retro-store control automatically manage track ID entries and coordinate with non-volatile storage control for track allocation and data commitment. This automated self-management reduces the need for complex external coordination and simplifies the overall device architecture while maintaining data integrity.
Solution Approach 2:
The patent applies universality by using the existing retro-store control mechanism (originally designed for other purposes) to also manage track ID entries and coordinate failover data protection. This multi-functional use of existing controls reduces device complexity by leveraging existing infrastructure rather than introducing entirely new complex subsystems.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
In a data storage subsystem with disk storage and a pair of clusters, one set of DASD fast write data is in cache of one cluster and in non-volatile data storage of the other. In response to a failover of one of the pair of clusters to a local cluster, the local cluster converts the DASDfast write data in local cache to converted fast write data to prioritize the converted data for destaging to disk storage. In response to failure to destage, the local cluster allocates local non-volatile storage tracks and emulates a host adapter to store the converted fast write data by the local non-volatile storage, reconverting the converted fast write data of the non-volatile storage to local DASD fast write data stored in the local non-volatile storage and stored in the local cache storage.