The Design and Implementation of a Capacity-Variant Storage System
summary
A capacity-variant storage system (CVSS) is implemented atop Linux 5.15 with CV-FS (F2FS baseline), CV-SSD and a user-space CV-manager that coordinates capacity shrinkage; CV-SSD classifies blocks as young / middle / edge and maps out edge blocks early, forgoing global wear leveling. CV-FS uses a remap vendor-unique command to reassign LBAs without rewriting data. Evaluation with synthetic and real workloads shows reduced latency, 49–316% higher throughput and 268–327% longer lifetime while trading away exported capacity (latency gains reported 8–53%).
AI-written summary · about
chapters (12)
- 0:00Introduction
- 0:38SSD aging impact
- 3:48Design principles
- 5:35File system elastic capacity
- 8:50Low-overhead capacity shrink
- 10:36Device-level consistent performance
- 13:50Block allocation and GC policies
- 14:56Implementation setup
- 16:20Synthetic workload performance
- 17:44Real workload performance
- 18:48Lifetime under constraints
- 19:51Summary
related talks
31mDesign and implementation of Autocaching for CXLSSD
15mD2FS: Device-Driven Filesystem Garbage Collection
14mZNS+: Advanced Zoned Namespace Interface for Supporting In-Storage Zone Compaction
36mFlash Friendly File System
24mSSDFS: ZNS SSD ready file system with zero GC overhead
17mHow advancements in storage are triggering innovations in open-source and vice-versa
25mShared CXL 3 memory what will be required
16mRearchitecting Buffered I/O in the Era of High-Bandwidth SSDs