Computational Storage Garbage Collection With In-Place Data Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Storage systems face performance degradation due to limited space availability, necessitating urgent garbage collection that increases I/O operations, which can impact performance, especially when slack space is depleted.
Innovation Solution
Leveraging computational storage capabilities to perform format conversion and compression during garbage collection, allowing for opportunistic space recovery by converting data to a more compressed or lossy format, thereby reallocating space efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If traditional garbage collection is performed in storage systems, then free capacity is reclaimed, but I/O operations increase and performance degrades
Solution Approach 1:
The patent extracts the garbage collection function from the storage controller and relocates it to the computational storage device. This separation allows the storage controller to focus on I/O operations while the computational storage device handles space reclamation independently, reducing the performance impact on the main storage system.
Solution Approach 2:
The patent introduces an intermediary mechanism where metadata about transformable data is stored and processed. This metadata acts as a mediator that enables the system to identify and compress transformable data during garbage collection, allowing for more efficient space reclamation without proportionally increasing I/O operations.
2Quantity of substance
If data is compressed during garbage collection, then storage capacity is optimized, but additional processing time is required
Solution Approach 1:
The patent applies preliminary action by pre-identifying transformable data through metadata tagging before garbage collection occurs. This allows the system to prepare a list of compressible data blocks in advance, so that during garbage collection, only the identified blocks need to be processed, reducing the overall processing time required.
Solution Approach 2:
The patent changes the parameter of data representation by converting data to different formats (e.g., from raw to compressed formats) based on transformability metadata. This parameter change enables more efficient storage utilization while the computational storage device handles the transformation during garbage collection operations.
3Quantity of substance
If computational storage is used to perform format conversion, then space reclamation is improved, but device complexity increases
Solution Approach 1:
The patent applies universality by designing the computational storage device to perform multiple functions: traditional storage operations, garbage collection, and format conversion/compression. This multi-functionality consolidates what would otherwise require separate systems into a single device, managing complexity while improving space reclamation capabilities.
Solution Approach 2:
The computational storage device performs self-service by autonomously executing garbage collection and format conversion operations based on metadata it processes itself. This self-service capability reduces the burden on the storage controller and eliminates the need for external processing systems, managing system complexity while achieving improved space reclamation.
Data Source
AI summary
Metadata associated with a file is received, where the metadata indicates whether data included in the file is transformable or not. The metadata of the file is processed to determine that the metadata indicates that the data is transformable. The data is transformed in accordance with indication provided in the metadata during garbage collection in a computational storage.