Backblaze Publishes its First SSD Reliability Report
Cloud storage provider Backblaze is known for its quarterly hard drive reliability reports, which are based on the failure rates of the hundreds of thousands of hard drives it has spinning away in its storage pods. Like most companies that run data centers, Backblaze’s primary storage needs are still handled by conventional hard drives. At the end of 2018, however, the company started adding SSDs to its storage ecosystem to serve as boot drives for its HDD-equipped servers.
Now that the company has several years of data to pore over regarding its SSDs’ performance, it has released the first reliability report for the handful of drives it’s been using. Though there are some outliers, and a few caveats, the bottom line is almost all of its SSDs have been extremely reliable. Overall, Backblaze is using 2,200 SSDs. While that’s a decent number of drives, it’s just 10 percent the size of the company’s HDD deployment. The sample size is therefore somewhat limited. Over time as it adds more drives and acquires more data on its existing drives, a clearer picture will come into focus.
For the entire report is that Backblaze’s cumulative Annualized Failure Rate (AFR) for all SSD drives as of Q4 2021 was 1.07 percent. That is just a bit higher than the AFR in 2021 for its 200,000+ hard drive pool, at 1.01 percent. As a reminder, annualized failure rate is not the same as “annual” failure rate, as the company takes into account the number of days the drive in question has been active when considering its overall reliability. Many drives are swapped out or replaced instead of just being installed and then operating indefinitely.
Also, although Backblaze’s hard drive pool is relatively diverse with various makes, models, and capacities, its SSD army is much smaller, comprising only 10 models from four brands. There’s six Seagate drives, two from Micron, and one drive from Dell and Crucial. Also, if you were hoping to see some “official” numbers from Backblaze for the vaunted Samsung SSDs, the report notes that it actually installed 10 850 EVO 1TB drives as the very first SSDs in its systems in order to test using them as boot drives, as it had never done this before. After the pilot program was successful, they were replaced after two weeks and “deployed to their original purpose,” which probably means staffers took them home.
Easily the most unreliable drive in their pool is the 500GB Crucial CT250MX500SSD1, which suffered a crazy AFR of 43.22 percent. However, they only deployed 80 of them and for a very short duration of just 0.4 months, so Backblaze says its needs “a lot more data” on the drives’ longevity before passing judgement. The 2TB Seagate ZA2000CM10002 also displayed a poor 28.81 percent AFR, but again, that was with just three drives in use. It’s not a large enough sample size to draw definitive conclusions.
Only one of its drive suffered zero failures, the 500GB Seagate ZA500CM10002. Since there were only 18 of them deployed, Backblaze’s confidence rating on their reliability is a too-high 6.7 percent. In order for the company to think a rating is reliable it needs a confidence interval of around one percent, and preferably closer to 0.6 percent.
The big takeaway here is similar to what we have reported previously, which is that SSDs and HDDs are more similar than people think when it comes to reliability, and SSDs are not that much more reliable despite having no moving parts. As Backblaze notes in its SSD report, the AFR for all of its SSDs is 1.07 percent, and the lifetime AFR for all of its hard drives is just 1.40 percent. However, Backblaze states that right now its SSD pool and hard drive pools are at vastly different phases of their lifecycles. Much more data is required to deliver more accurate findings. As always, Backblaze says it will continue to monitor its solid-state drives and continue to add to these findings in future months.