LBZ: A Lightweight Block Device for Supporting F2FS on ZNS SSD

Zoned NameSpace SSD (ZNS) is emerging as a new paradigm for serving log-structured storage system (e.g. F2FS). However, existing ZNS SSD cannot directly support random write, which is still highly required by the metadata region of log-structured storage system. In this paper, we propose LBZ, a lightweight in-kernel block device running on ZND SSD with the log-structured design to allow random writes. Specifically, we design LBZ to hold the metadata area of F2FS such that F2FS can run on single ZNS SSD without using extra conventional SSD or conventional namespace of ZND SSD. LBZ first utilizes the out-of-band area of NAND flash chips to maintain the mappings between logical blocks and physical zone offsets. Since logical blocks of LBZ are sequentially written into ZNS SSD, LBZ incurs garbage collection overhead when logical blocks become invalid. In order to support efficient garbage collection in LBZ, we then propose multi-stream separation to distinguish different metadata areas of F2FS and place metadata with similar lifetimes into same zones. Meanwhile, LBZ executes three-phase checkpoint to get aware of invalid blocks and reclaim the blocks itself without modifying F2FS. We implement LBZ on real ZNS SSD and evaluate its basic performance as well as its garbage collection efficiency. The experiment results show LBZ can fully exploit the bandwidth of ZNS SSD. Moreover, LBZ provides efficient garbage collection by reducing the number of written blocks by up to 19x.

Authors

Institutions

Publication Details

Published
2024-11-18
DOI
https://doi.org/10.1109/iccd63220.2024.00059
Citations
1
Primary Topic
Advanced Data Storage Technologies
Type
article
Field-Weighted Citation Impact
0.35
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

LBZ: A Lightweight Block Device for Supporting F2FS on ZNS SSD

Peng Liang, Hao-Chiang Hsu, Bo Jiang, Zifeng Yang et al.
1 citations
Advanced Data Storage Technologies
0.35
article

LBZ: A Lightweight Block Device for Supporting F2FS on ZNS SSD

Peng Liang, Hao-Chiang Hsu, Bo Jiang, Zifeng Yang, Dejun Jiang, Yongpeng Yang
article en
1 citations

Abstract

Zoned NameSpace SSD (ZNS) is emerging as a new paradigm for serving log-structured storage system (e.g. F2FS). However, existing ZNS SSD cannot directly support random write, which is still highly required by the metadata region of log-structured storage system. In this paper, we propose LBZ, a lightweight in-kernel block device running on ZND SSD with the log-structured design to allow random writes. Specifically, we design LBZ to hold the metadata area of F2FS such that F2FS can run on single ZNS SSD without using extra conventional SSD or conventional namespace of ZND SSD. LBZ first utilizes the out-of-band area of NAND flash chips to maintain the mappings between logical blocks and physical zone offsets. Since logical blocks of LBZ are sequentially written into ZNS SSD, LBZ incurs garbage collection overhead when logical blocks become invalid. In order to support efficient garbage collection in LBZ, we then propose multi-stream separation to distinguish different metadata areas of F2FS and place metadata with similar lifetimes into same zones. Meanwhile, LBZ executes three-phase checkpoint to get aware of invalid blocks and reclaim the blocks itself without modifying F2FS. We implement LBZ on real ZNS SSD and evaluate its basic performance as well as its garbage collection efficiency. The experiment results show LBZ can fully exploit the bandwidth of ZNS SSD. Moreover, LBZ provides efficient garbage collection by reducing the number of written blocks by up to 19x.

Chinese Academy of Sciences (CN), Institute of Computing Technology (CN)
Openalex Percentile: Top 35%
Advanced Data Storage Technologies
0.35
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.