Metadata Operations in Persistent Memory Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing file systems face performance degradation and reliability issues when handling metadata operations due to the append-only mode of ZNS SSDs, which restricts write operations and introduces overhead in read I/O requests, and journal replay may not significantly improve performance without computational power in storage devices.
Innovation Solution
Implementing a replicated metadata model using persistent and byte-addressable memory on the host side, where the file system driver stores metadata modifications in a journal and periodically flushes them to the storage device, ensuring high performance and reliability, and utilizing computational storage devices for journal replay to maintain metadata integrity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If journal replay is implemented to improve metadata reliability, then metadata integrity is improved, but read I/O performance deteriorates due to additional overhead
Solution Approach 1:
The patent implements preliminary action by pre-computing and caching metadata snapshots before they are needed. The system periodically creates consistent snapshots of metadata state and stores them in advance, so when read I/O requests arrive, the data is already prepared and available immediately, eliminating the performance penalty of on-demand journal replay.
Solution Approach 2:
The patent applies copying by creating duplicate copies of metadata in snapshot form. Instead of replaying the entire journal for every read operation, the system maintains copies of metadata at specific points in time, allowing rapid access to historical or consistent states without reprocessing the journal.
2Reliability
If append-only mode is used on ZNS SSD to ensure write reliability, then write integrity is improved, but write performance deteriorates due to restricted write operations
Solution Approach 1:
The patent applies segmentation by dividing metadata storage into two distinct segments: a write-optimized journal area that accepts append-only operations for high write performance, and a snapshot area that stores periodic consistent copies for reliability. This segmentation allows each segment to be optimized for its specific purpose without compromising the other.
Solution Approach 2:
The patent introduces snapshots as an intermediary between the append-only journal and the final reliable storage state. Rather than requiring every write to be immediately verified and committed to the final storage location, writes are first recorded in the journal, and snapshots serve as intermediate checkpoints that guarantee consistency, thereby improving overall write performance while maintaining integrity.
3Speed
If journal replay is offloaded to storage device to improve metadata operation speed, then processing speed is improved, but device complexity increases due to computational requirements
Solution Approach 1:
The patent applies preliminary action by performing metadata snapshot creation during idle periods or in the background, rather than computing everything on-demand. The storage device prepares metadata snapshots in advance during low-utilization times, so when fast metadata operations are needed, the pre-computed snapshots are immediately available, achieving high speed without requiring complex real-time computation capabilities.
Data Source
AI summary
The present disclosure describes techniques for implementing fast and reliable metadata operations. A metadata area instance may be created in a persistent memory associated with a host. The metadata area instance may comprise a first portion configured to store an initial state of metadata, a second portion configured to store an actual state of the metadata, and a third portion configured to store a plurality of modifications to the metadata. A main copy of the metadata may be generated by performing write operations in the metadata area instance. The main copy of the metadata may be updated based on receiving information indicative of a modification to the metadata.