Which is the worst-case scenario: dying HDD or SSD?

Question

We all know SSD offer many advantages (such as speed) over HDD drives but which one is safer when it comes to data rescue if, for instance, the drive is failing?

From my understandings on SSD drive data is stored on-chip and when chip dies all data is lost (like on USB flash drive), but on HDD you can detect pre-failing and prevent data loss.

Is this true? In a search for expert opinion on this. Thanks.

Answer 1

Disclaimer: when a storage unit/subsystem fails, it is way better to have valid backups rather than try the recovery lottery.

That said, while modern SSDs have lower failure rate than mechanical HDDs, when they fail, they fail catastrophically. Rather than individual NAND chips, quite often is the FLT (Flash Translation Layer) or the controller which fails. In this case, recovery is extremely hard, as basically all SSDs scramble data as they are written on the NAND chips. This means that even having expensive laboratory equipment to de-solder the BGA chips, a direct reading from them is going to be relatively useless. Rather, it is a first step only to a much harder/longer recovery process.

On the other side, HDD are relatively "simpler" beast. Even the most damaged HDD (short of total disk disruption) can be read by carefully transplanting the platter onto an appropriate device to do a bit-per-bit read of the magnetic data. Note, however, that extracting consistent data is another matter - for example, CoW filesystems (as ZFS and BTRFS) themselves scramble data at an higher (ie: logical) level.

Answer 2

It's complex question. Both have negatives, but different.

HDD:

• electronics can fail, but disk platters can survive and be put into different disk of the same type
• but also there can happen, that either some rubbish get's inside (can be some part of the disk itself) or the head touches the surface of disk platters and all (or most) content of the disk is gone.

SSD:

• controller may fail and data reconstruction from flash chips is very difficult or impossible
• some of flash modules fails and most of the disk is unreadable as data are stored almost evenly over all memory modules. But the disk may continue working, just with limited capacity (and lower performance)
• the disk hits its wearout limit (overwrites are no longer possible), however written data can be still red. So the SSD works as RO device. That means the data can be saved to different device intact (or almost intact). But the newer technology was used the sooner the information disappears from flash cells - so the recovery must be done ASAP.
• the SSD was left without electricity for some time (month or few) -> the controller can't autorefresh cell content and data slowly dissapears. So when you enable the disk after long time, you may end up with shiny clear disk, which is actually working. Again, the smaller technology is used, the sooner this efect comes.

So nothing's perfect...

Feel free to add additional scenarios.

Answer 3

Neither. Lack of backups is the worst case scenario with both. Data rescue should not be a plan against failing drives, it's really the last option. Therefore, you should choose between HDD and SSD for other reasons.