Wednesday, 2 February 2022

Backblaze Publishes Stats for Hard Drive Failure in 2021

The Backblaze 4U storage pods can save up to 480 TB of user data.

Cloud backup provider Backblaze has collected all the data about which hard drives failed in its pods in 2021, providing a keen insight into what it sees are the most — and least — reliable hard drives. Backblaze typically does this in a quarterly fashion, but since the year ended it decided to an annual wrap up of sorts for its hard drive army, and it is even sharing data from past years to see how things have changed over time. The good news: most of them are pretty dang reliable, regardless of capacity or manufacturer. Note that this report does not include information about SSDs, which will be published soon in a separate blog post.

Right off the top we should note Backblaze is reporting Annualized Failure Rate, not annual failure rate. This is due to the fact that drives come and go from Backblaze’s pods all the time so it has to calculate how long a drive has been in service to come up with a number for that specific drive’s failure rate. As an example it notes that a single drive might contributed to the pod for 43 days, 186 days, or all 365 days of 2021. To get this number Backblaze uses the following equation: AFR = (drive failures / (drive days / 365)) * 100

The major takeaway from the analysis is that the annualized failure rate for all 203,759 hard drive it examined was just 1.01 percent, which is a slight increase from 0.93 percent it recorded in 2020. However, in 2019 the annualized failure rate was 1.83 percent, so the failure rate dropped significantly last year, and in 2021 remained quite low; that’s good news for data archivers.

Backblaze says there is one reason for continued low failure rate throughout its drive farm: high capacity hard drives. It defines this as a 12TB, 14TB, or 16TB drive, and all of those drives had a collective failure rate below the 1.01 percent average, while the 4/8/10TB drives all failed at higher rates than average, with the 10TB drives dying the most often at 2.26 percent. Backblaze notes that one of the reasons its high capacity drives are more reliable than average is partly due to the fact that they are the newest drives to be added to the collection.

Comparison chart showing failure rate for both high and lower capacity hard drives. (Image: Backblaze)

One interesting tidbit from Backblaze’s reporting is the most reliable drive in the entirety of Backblaze’s storage array is also the oldest model in its fleet: the Seagate 6TB, model ST6000DX000. This particular drive has had a failure rate of just 0.11 percent, and an average lifespan of 80.4 months, so just a bit under seven years in service. It notes that it only has 886 of them in service, but that they are definitely “thumbing their nose at the bathroom curve,” which we discussed in a previous report on SSD and HDD reliability.

Another drive to watch out for is the 14TB Seagate, model: ST14000NM0138. Backblaze writes that in its previous update it noted the drives were failing at an alarming rate, and nobody could figure out why. After consulting with Seagate, and Dell (the drives were in Dell storage servers) it was decided to update the firmware of the drives, which reduced its failure rate from 6.29 percent in Q3, to 4.66 percent in Q4; a meaningful improvement. Still, those numbers are way higher than any other single model Backblaze uses, but these appear to be drives being used for some kind of in-office storage as it says 19 of them failed in Q4, and they usually count drives in the hundreds if not thousands. It could be an anomaly, however, and Backblaze says it is continuing to investigate.

Annualized failure rate by manufacturer. (Image: Backblaze)

As far as which drive manufacturer is the most reliable, Backblaze uses four companies: Western Digital (WDC), HGST, Toshiba, and Seagate. Of the four Seagate has the highest failure rates, but the number is still quite low overall, and the results of the chart below reinforce the most salient point of the company’s data collection: hard drives are very reliable, across brands, and of varying sizes. For example, the worst failure rate reported for Seagate over the the past three years was 2.7 percent in Q4 2019, and in Q4 2021 it was just 1.48 percent, so things improved. HGST is the most reliable, which has been a recurring theme in Backblaze’s quarterly updates, as over the past three years the company never rose above one percent failure rate at any time, and finished the year with an impressive 0.37 percent failure rate in Q4 2021.

Now Read:



No comments:

Post a Comment