NVM Update Layer Retention to Reduce Consolidation Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data storage systems face inefficiencies in consolidation processes after sequential writes, leading to a high number of back-to-back consolidations and performance degradation due to immediate erasure of update layer entries, which affects read performance and increases latency.
Innovation Solution
Implement a delayed copy-back mechanism where update layer entries are not immediately erased after consolidation, allowing some data to remain in the update layer for future read operations, and utilize a bitmap to track these entries for efficient space management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data in the update layer is immediately erased after consolidation, then storage space is freed efficiently, but read performance deteriorates due to frequent re-copying operations
Solution Approach 1:
The patent applies preliminary action by maintaining data in the update layer before consolidation is complete. The controller keeps a copy of data in the update layer and only erases it after confirming that the master layer has been successfully updated, allowing read operations to access the data without requiring immediate re-copying. This preliminary retention of data resolves the contradiction by ensuring read performance while still achieving space efficiency.
2Quantity of substance
If all data in the update layer is erased after consolidation, then space is maximized for new data, but unnecessary consolidations increase latency
Solution Approach 1:
The patent applies partial action by selectively erasing only the data in the update layer that has been successfully consolidated to the master layer, while retaining data that may still be needed for read operations. The controller tracks which data has been consolidated and erases only that portion, avoiding unnecessary erasure of data that might be read soon. This partial erasure approach reduces unnecessary consolidations and latency while still maximizing available space.
3Reliability
If the update layer maintains all data for read operations, then read performance is improved, but the number of back-to-back consolidations increases
Solution Approach 1:
The patent applies feedback by implementing a tracking mechanism that monitors which data in the update layer has been consolidated to the master layer. The controller uses this feedback information to determine which data can be safely erased and which should be retained. This feedback loop allows the system to maintain read performance by keeping necessary data while reducing consolidation frequency by erasing only consolidated data, thus resolving the contradiction between read performance and consolidation frequency.
Data Source
AI summary
When performing a consolidation process, entries (uRuns) in the update region (uRegion) of an update layer (uLayer) are typically erased immediately after the data in the uRuns is updated to a master layer (mLayer). The data may, at a later time, be copied back to the uRuns for read purposes. Rather than erasing the uRuns and then copying the data back to the uRuns for read purposes, the uRuns can simply not be erased after the consolidation. If space is needed, the uRuns can be erased without any need for consolidation. Furthermore, if another consolidation process occurs, the uRuns that were previously consolidated, but not erased, can simply be erased.