Backblaze, a provider of cloud storage, published the stats (opens in new tab) For 2,906 SSDs used as boot drives for storage servers.To clarify, these are not that’s all The boot drive also reads and writes log files and temporary files, so the former can cause significant wear and tear. Backblaze has been using SSDs for its boot drives since 2018, and similar to hard drive stats, most are among the few that give as much insight into how massive amounts of consumer drive capacity hold up over time. One way.
Before we get into the stats, there are a few conditions. First, most of the SSDs Backblaze uses aren’t the latest M.2 NVMe models. They are also generally very small in capacity, with most drives only offering 250 GB of storage, and about a third more with 500 GB, and only three with larger 2 TB drives. However, when it comes to managing hardware, having lots of the same model hardware simplifies things. Anyway, if you look at popular drive stats and you might end up on the best SSD list, you’ll be disappointed.
Here are last year’s stats.
The 7 drive model used did not fail, but only one of the many drives installed is the Dell DELLBOSS VD (Boot Optimized Storage Solution). The other 6 are using less than 40 SSDs and 4 are only installed on 2 or 3 servers. It is more beneficial to pay attention to heavily used drives, especially those with over 100,000 drive days.
These consist of a Crucial MX500 250GB, Seagate BarraCuda SSD and SSD 120, and Dell BOSS. The average age of the Crucial MX500 is (opens in new tab) The MX500 was first available in early 2018, but the drive is only 7 months old. It’s clear that Backblaze wasn’t an early adopter of his latest SSD. Still, overall, boot SSDs have an annual failure rate of less than 1%.
Going back to a longer-term perspective over the past three years, we have maintained an annualized failure rate of 1%, with a total of 46 drive failures during that period. Backblaze also notes that the MX500 improved significantly for his 2022 after two relatively quick breakdowns in 2021. Seagate’s old ZA250CM10002 failure rate also dropped to 2% last year, but his newer ZA250CM10003 had longer days and fewer failures, so it will continue to do so. It will be interesting to see if these trends continue.
Another piece of data Backblaze looked at is the SSD temperature reported by SMART (Dell BOSS doesn’t seem to support this). While the chart is not zero-based, it may appear to have a lot of variability, but in reality the drive ranged from 34.4C to 35.4C on average — just 1C span.
Of course, this is just an average and there are outliers. A drive of 20C he observed four times and a drive of 61C once, mostly in the range of 25-42 degrees Celsius. It would be nice if the bell curve above somehow correlated with failed drives, but with a total of 25 failures per year, it didn’t — Backblaze called the resulting plot “nonsense.” I called.
Ultimately, the number of SSDs used in Backblaze pales in comparison to the number of hard drives — check out the latest Backblaze HDD report (opens in new tab)For example, over 290,000 drives were in use in the past year. This is because customer data is not stored on his SSD. In other words, the SSD is dedicated to OS, temp files, and logs. Still, data from about 3,000 drives is far more than we (non-IT guys) are likely to access in his year. By the way, the HDD’s AFR was 1.37%.
Does this prove that SSDs are more reliable than HDDs? No, it’s still important to have a good backup strategy. Hopefully, in the next few years, more recent M.2 SSDs will be incorporated into Backblaze’s data. It is assumed that the storage server supports these interfaces. Given the time, it almost certainly will.
