hesela.devSilent data corruption
Silent data corruption is a data-integrity error that a specified storage component or layer does not detect or report.
State the observation boundary. A drive can return incorrect data without reporting an error, while a higher-level checksum or identity check detects it. The FAST 2008 field study used silent for corruption missed by the drive and other hardware, not necessarily by the whole system. [1]
A latent sector error that produces an explicit error on access is a different observation. [2] A matching checksum can accompany stale data after a lost write; detection and successful recovery are separate outcomes. [3]
In the paper's single-error RAID model, a misdirected block can carry a higher version number. Physical identity is checked before versions to avoid selecting that wrong block for repair. [3]
Relations
Citations
- Bairavasundaram et al., An Analysis of Data Corruption in the Storage Stack, FAST 2008, sections 2.2-2.3.
- Bairavasundaram et al., An Analysis of Latent Sector Errors in Disk Drives, SIGMETRICS 2007, sections 2.2 and 3.2-3.3.
- Krioukov et al., Parity Lost and Parity Regained, FAST 2008, sections 2.3, 4.4 and 4.6.
Further reading
Sources & caveats
- The definition specifies an observation boundary; it is a synthesis, not a quotation from a universal storage standard. Historical detected events do not estimate today's undetected-error rate.