Host Entity Cache Synchronization for Storage Data Integrity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Operating storage virtualization systems in write-back caching mode can provide performance benefits but increases the risk of data loss due to power failures or system malfunctions, whereas write-through caching ensures data security at the cost of performance.
Innovation Solution
Implementing a Completion Pending PSD Cache Synchronization list to manage write IO requests, where the host entity delays clearing cache data and issues a synchronized cache IO request to ensure data is securely committed to physical media, allowing for performance similar to write-back caching while maintaining data security akin to write-through caching.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If write-back caching mode is used, then IO processing performance is improved, but data security deteriorates due to risk of data loss during power failures or system malfunctions
Solution Approach 1:
The system performs preliminary actions by maintaining a completion pending list that tracks all write IO requests before they are fully processed. This allows the system to prepare for potential data loss scenarios by having a record of all uncached writes, enabling subsequent synchronization operations to ensure data safety without compromising the performance benefits of write-back caching.
Solution Approach 2:
The system implements feedback mechanisms through the synchronized cache IO request that queries the completion pending list after power restoration or system recovery. This feedback loop ensures that any data that was cached but not yet committed is identified and re-committed, providing assurance that data security is maintained while allowing write-back caching to operate for performance optimization.
2Reliability
If write-through caching mode is used, then data security is ensured, but IO processing performance deteriorates
Solution Approach 1:
The system segments the data safety assurance process into two distinct phases: the normal write-back caching operation for performance, and a separate synchronization phase triggered by power restoration or system recovery events. This segmentation allows the system to enjoy the performance benefits of write-back caching while periodically ensuring data security through targeted synchronization operations, rather than continuously using write-through mode.
3Productivity
If cache data is cleared immediately after successful write, then processing efficiency is improved, but data integrity deteriorates during power failures
Solution Approach 1:
The system performs preliminary action by adding write IO requests to the completion pending list immediately upon receiving their completion responses, before any cache clearing occurs. This preliminary tracking ensures that when power failure or system malfunction occurs, the system has already recorded all uncached writes and can subsequently synchronize them to maintain data integrity while still allowing efficient processing during normal operation.
Data Source
AI summary
A method for committing cached data to storage media including a host entity and a physical storage device (PSD), comprising: the host entity issuing a write IO request to the PSD being in write-back caching mode to write data from a host memory to the PSD; the host entity receiving a successful IO completion response associated with the write IO request from the PSD and recording an entry of the write IO request to a selected list; the host entity issuing a synchronized cache IO request to the PSD; the PSD responding a successful IO completion response of the synchronized cache IO request to the host entity; the host entity extracting the entry of the write IO request from the selected list before issuing the synchronized cache IO request; and the host entity engaging completion processing of write IO request corresponding to the entry of the write IO request extracted.