A Study of SSD Reliability in Large Scale Enterprise Storage Deployments
summary
Field study analyzes 1.4 million SSDs from NetApp filers over 2.5 years (3 makers, 18 models, 12 capacities, SLC/cMLC/eMLC/3D‑TLC) and reports annual replacement rates between about 0.7%–1.2% (much lower than typical HDDs). Key findings: firmware changes cause up to 10× differences in replacement rates, infant mortality can last over a year before stabilizing, only ~2% of RAID groups see a failure but 46% of follow‑up failures occur the same day and ~2.54% follow within a week.
AI-written summary · about
chapters (13)
- 0:00Introduction to SSD study
- 0:46Scope and methodology
- 2:00Data set characteristics
- 3:23Replacement types and rates
- 5:06Field reliability results
- 6:36Impact of flash type
- 8:38Lithography and density
- 10:16Usage and wear factors
- 11:38Drive age and failure rates
- 13:31Firmware impact
- 15:29Raid group correlations
- 18:04Conclusions and takeaways
- 20:22Question and answer session
related talks
09mTest of Time Award: Analysis of Data Corruption in the Storage Stack
16mPerseus: A Fail-Slow Detection Framework for Cloud Storage Systems
49mThe Math of Reliability
15mShit, our Site is Down! Let's build a culture of reliability
37mRelia...bility?
16mEngineering Failure-Resilient Systems
32mThe culture of Site Reliability Engineering
51mShrinking Production Incidents