Maximizing SD Card Life, Performance, and Monitoring with KrillKounter
summary
Empirical tests use fio on Raspberry Pi Pico Z2 clusters to repeatedly write block-layer data to Transcend 8GB MLC SD cards and measure failure, showing a sequential 512KB workload reaches about 40 TB written (vs an estimated 21 TBW) before errors (CRC, -110 timeouts) and read-only or disappearing-device failures. Practical fixes include raising average write size, enable discard/TRIM, increase filesystem cluster size, use mq-deadline with larger nr_requests, buy larger/TBW or A1-profile cards, and run Krill Kounter to log block stats for lifetime monitoring.
AI-written summary · about
chapters (23)
- 0:00SD card reliability issues
- 1:24SD card construction
- 3:13NAND flash fundamentals
- 6:45Floating gate transistors
- 10:16NAND write amplification
- 13:47NAND flash degradation
- 15:33Wear leveling for lifespan
- 17:20Controller firmware limits
- 18:23Endurance testing setup
- 21:16Endurance test results
- 23:40How SD cards fail
- 25:47Key endurance findings
- 27:13Optimizing SD usage
- 31:02Reducing write amplification
- 32:28Performance bottlenecks
- 33:33Evaluating SD classes
- 35:41Introducing KrillKounter
- 36:44Key takeaways
- 37:13Q&A session part one
- 39:06Q&A session part two
- 41:00Q&A session part three
- 43:28Q&A session part four
- 45:52Closing remarks
related talks
36mLet's Play Mini-card-sized Computer Boards on the Business!
54mBetter Alignment of Flash Storage to Mobile System Behavior
36mLet's Play Mini-card-sized Computer Boards on the Business!
01h 01mBoosting Software Efficiency : A Case Study of 100% Performance Improvement in ...
20mSmartDedup: Optimizing Deduplication for Resource-constrained Devices
36mFrom Raw To Refined: The Evolution of Raw Flash Support in Linux
01h 01mDiskless Booting of RPi 400 Using Bootp, tftp, iscsi and NFS
14mModernizing File System through In-Storage Indexing