SSDs are more reliable than HDDs but you need to know more

Whenever there is discussion about SSDs versus HDDs, the focus seems to be on the higher IOPS, transfer rates, and lower power consumption of SSDs. We seldom discuss reliability. The absence of an industry consensus and a lack of published data (with the exception of Backblaze, which regularly shares insights on their data center's hard drives and SSDs) doesn't help either.

For now , the only available data comes from Backblaze, and their findings suggest a lower failure rate for SSDs compared to hard drives. You can read the entire article here. However In my opinion it's important to approach this report with a nuanced perspective, and not interpret it as SSDs being universally more reliable than hard drives. Read on if you want to have an alternative viewpoint about reliability of SSD vs HDD

First - SSDs and HDDs in the Backblaze report are not under the same workload

Backblaze utilizes hard drives (data drives) for storing data and employs SSDs for booting purposes (boot drives). It's important to recognize that SSDs don't store the same type of data as HDDs.

SSDs are more prone to wearing from writing as they have a TBW rating (Terabytes Written) as their flash cells have limited P/E Cycles. Consequently, claiming definitively that SSDs are more reliable is challenging unless they undergo a comparable read/write workload as hard drives.

Second- Data Recovery and Early Warnings are more important than longer time to failure

What holds greater significance than an 'extended time to failure' is your ability to recover from such failures and having ample warning time for a timely response.

While data recovery is possible for failed SSDs, the success rates are comparatively lower than for HDDs. Data recovery on SSDs that didn't fail ( deleted data recovery for example) also is lower on SSDs due to TRIM. So even if your HDD has a higher chance of failure, you have a higher chance of recovering your data from HDD also.

Equally important is the manner in which a device fails- definitely more important than the longer time to failure.

For instance, in a RAID configuration with 5 drives (RAID 5 with 1 HDD failure tolerance), the concern lies more in the drives failing at different times than their individual reliability.

If two drives fail simultaneously after 10 years, it poses a significant challenge since your RAID could survive only 1 failure. However, if one drive fails after 1 year, allowing time for replacement, RAID rebuilding, and another drive fails after 3 years, your RAID survives . In the second case even though your drives were 5 times less reliable, they failed at different times , giving you time to correct their failure and it was more important than having a longer time to fail.

While I lack specific data on the matter, logic dictates that the likelihood of two mechanical drives of the same age failing simultaneously is lower compared to the probability of two equally aged SSDs failing concurrently.

I trust this provides you with an alternative viewpoint on the Reliability of SSDs vs. HDDs.