OPTR: Order-Preserving Translation and Recovery Design for SSDs with a Standard Block Device Interface
summary
Proposes OPTR, an FTL+recovery design that keeps the standard block device interface and gives three request‑level crash guarantees: ordered durability, per‑request atomicity, and flush semantics. Realized on real SSD hardware and evaluated by optimizing Ext4 and SQLite; shows 1.27× (Ext4), 2.85× (Ext4+SQLite) and 6.03× (no‑barrier) speedups. Uses write completion tracking (write ID + size per page), a flow‑network cut to pick recoverable writes, supports coalescing with coalescing records, and requires scanning pages only after the latest checkpoint.
AI-written summary · about
chapters (13)
- 0:00Start
- 1:16Traditional SSD: Weak Crash Guarantees
- 1:47OP-SSD: Strong Crash Guarantees
- 2:45Outline
- 3:51Background: GC and SSD Recovery
- 5:17Order-Preserving Recovery
- 6:27Recovery Procedure (without write cache and GC)
- 10:45Recovery Procedure (with write cache and GC)
- 11:53System Optimizations and Evaluation
- 12:31Baseline Systems
- 12:53Systems Using Transactional SSDs
- 13:133rd System Optimization with OP-SSDs
- 14:20Conclusion
related talks
13mRemap-SSD: Safely and Efficiently Exploiting SSD Address Remapping to Eliminate...
53mOvercome Blast Radius Anxiety With Ultrafast Rebuilds and Performance for High-capacity SSDs
12mFusionRAID: Achieving Consistent Low Latency for Commodity SSD Arrays
30mHybrid XFS: Idea to Exabytes in 12 Months
13mLiquid-State Drive: A Case for DNA Block Device for Enormous Data
01h 07mEmerging Embdedded Memories and Storage
49mSSDs for Your Database
45mBig Data and SSDs