A new study, recently published at FAST 13, has demonstrated that the majority of SSDs on the market suffer severe vulnerability to unclean shutdown/unexpected power outage scenarios.
The investigation has shown findings reminiscent of rumors from the early days of SSDs, where certain companies gained a reputation for poor quality, due to power cycle bugs. However, unlike relatively isolated cases attributed to one particular brand of controller, this new study shows that similar, albeit less frequent, bugs exist in most drives on sale today.
One drive model was found to be bricked by power outages while the drive was processing writes. Other models have shown total data corruption, and others have shown severe data corruption affecting approximately 30% of the data on the drive.
Most other drives tested corrupted or missed the writes in progress, most in ways which break the crash recovery algorithms in journaled file systems, databases, etc.
Even drives with supercapacitor power backup were found to corrupt data when power was last while the drive was busy.
Link
The investigation has shown findings reminiscent of rumors from the early days of SSDs, where certain companies gained a reputation for poor quality, due to power cycle bugs. However, unlike relatively isolated cases attributed to one particular brand of controller, this new study shows that similar, albeit less frequent, bugs exist in most drives on sale today.
One drive model was found to be bricked by power outages while the drive was processing writes. Other models have shown total data corruption, and others have shown severe data corruption affecting approximately 30% of the data on the drive.
Most other drives tested corrupted or missed the writes in progress, most in ways which break the crash recovery algorithms in journaled file systems, databases, etc.
Even drives with supercapacitor power backup were found to corrupt data when power was last while the drive was busy.
Link
Last edited: