SSDs aren't as reliable as we think
Cloud storage company Backblaze has released new data showing that the failure rates of its SSDs are nearly as high as those of hard drives (HDDs). In a recent blog post, Backblaze published their analysis of SSDs and HDDs, which is based on real usage data. The company employs SMART diagnostics to check the health of its drives and has said it doesn't know why its SSDs have such a high failure rate right now.
Backblaze defines a defective drive both when not it is working properly that when it is about to fail. To predict the latter case, the company uses the internal SMART statistics of the units, recording the rates of reading errors, the level of wear, the hours of ignition, the total count of program errors and more. To make the analysis more useful, Backblaze only considered the boot drives on its servers, instead of the primary storage drives. These have almost constant usage from server startup to reading, writing and deleting files, with very little downtime.
Since 2018, Backblaze has been using a combination of SSDs and hard drives for the drives. startup on its servers, making the company the ideal candidate for this type of test. In the first table, Backblaze shows SSD and HDD failure rates as of 2013. You can see that HDDs have a significantly higher rate than SSDs, making us think that SSDs are indeed much more reliable than HDDs, as we have been used to. to hear forever.
Credit: Backblaze However, some problems remain. Backblaze only started installing SSDs in 2018, but the company has hard drive health data dating back to 2013, which skews the results a bit. After taking into account the age of the units, we can see that the results have changed significantly. SSDs are not far behind hard drives in the failure rate, 1.05% vs 1.38%.
Credit: Backblaze Backblaze doesn't know why SSDs are failing so badly , but the data certainly shows that they are not as durable as previously thought. The company also points out that its hard drive failure rates increased significantly between 2018 and 2020 (before stabilizing in 2021) due to the age of the drives (most were installed around 2014). This means that SSDs could suffer the same fate as well, but only time will tell us if this will happen.
Credit: Backblaze For now, it's not clear why these SSDs fail so often, especially when they have no moving parts. The models used by Backblaze have not been disclosed - some may employ budget controllers and / or NAND flash, or perhaps other factors are at play. As the company gains more long-term data using SSDs, the picture will also become clearer. Whatever the reason for failures, surely it is always best to back up storage drives or use RAID arrays with redundancy, regardless of the technology used.
Backblaze defines a defective drive both when not it is working properly that when it is about to fail. To predict the latter case, the company uses the internal SMART statistics of the units, recording the rates of reading errors, the level of wear, the hours of ignition, the total count of program errors and more. To make the analysis more useful, Backblaze only considered the boot drives on its servers, instead of the primary storage drives. These have almost constant usage from server startup to reading, writing and deleting files, with very little downtime.
Since 2018, Backblaze has been using a combination of SSDs and hard drives for the drives. startup on its servers, making the company the ideal candidate for this type of test. In the first table, Backblaze shows SSD and HDD failure rates as of 2013. You can see that HDDs have a significantly higher rate than SSDs, making us think that SSDs are indeed much more reliable than HDDs, as we have been used to. to hear forever.
Credit: Backblaze However, some problems remain. Backblaze only started installing SSDs in 2018, but the company has hard drive health data dating back to 2013, which skews the results a bit. After taking into account the age of the units, we can see that the results have changed significantly. SSDs are not far behind hard drives in the failure rate, 1.05% vs 1.38%.
Credit: Backblaze Backblaze doesn't know why SSDs are failing so badly , but the data certainly shows that they are not as durable as previously thought. The company also points out that its hard drive failure rates increased significantly between 2018 and 2020 (before stabilizing in 2021) due to the age of the drives (most were installed around 2014). This means that SSDs could suffer the same fate as well, but only time will tell us if this will happen.
Credit: Backblaze For now, it's not clear why these SSDs fail so often, especially when they have no moving parts. The models used by Backblaze have not been disclosed - some may employ budget controllers and / or NAND flash, or perhaps other factors are at play. As the company gains more long-term data using SSDs, the picture will also become clearer. Whatever the reason for failures, surely it is always best to back up storage drives or use RAID arrays with redundancy, regardless of the technology used.